What is the lowest CR creature that can defeat an Ancient Red Dragon in open combat, and how many of them does it take?

Question

This Q&A explores a similar question with commoners specifically, but I wanted to pose a challenge that isn't defeated by kiting. As such, I figured I'd open the floor to all creatures, with a priority for lower CRs and fewer creatures (in that order).

Assume the battlefield does not contain cover and has sufficient space for the dragon to always use its full flight to make distance from its enemies.

The dragon employs the following tactics:

• Starts far enough away that the enemy can't close the gap on their first turn; the dragon uses Fire Breath on its first turn in which an enemy is within 90 feet, and then begins to kite.
• The dragon maintains a 350 foot altitude unless it needs to descend to score a Fire Breath.
• Uses the Dash action and Wing Attack to get distance from the horde (i.e. 200 feet per round).
• Once Fire Breath has recharged, allow the horde to close the gap to within 90 feet (including by lowering its elevation to 90 feet); enough to use the breath weapon and then fly away again (only gaining a distance of 170 feet on its turn; 210 after the Wing Attack which also knocks prone any nearby creatures that fail the save).
• If the creatures make ranged attacks when it goes for its Fire Breath turn, it flies forward to Fire Breath as many of them as possible before resuming kiting.
• Preserve Legendary Resistance for effects that hamper its movement.
• Use Tail Attack once when possible.

Please describe not only what creatures makes up your horde, but also the tactics the horde employs. A horde with the lowest CR is a good answer. If you can tie another answer, but with fewer creatures (assuming average results) that is a better answer.

Answer 1

At most 2500 CR 1/8 Harengon Brigands

The Harengon Brigand has a sling attack with range 30/120 that does +5/1d4+3 damage.

If it attacks with disadvantage, then it hits on a 17-20, so a 4% chance to hit. If we neglect critical hits then it does 5.5 damage on a hit, so 0.22 damage per attack.

The dragon has 546hp. We assume for simplicity that each brigand only attacks once before being killed by fire breath. (Modeling the actual number of brigands killed by fire breath is quite difficult!) An upper bound of 546/0.22=2482 brigands are needed.

---

It's possible to optimize these numbers quite a bit if we have a clear understanding of the dragon's tactics. We know:

If the creatures make ranged attacks when it goes for its Fire Breath turn, it flies forward to Fire Breath as many of them as possible before resuming kiting.

What does it mean to "fly forward to Fire Breath as many of them as possible"? Does this mean the dragon will actually land, adjacent to the army, to maximize the number caught in its cone? If so, then the brigands can use Pack Tactics and they all gain advantage, which increases their chance to hit from 4% (with 0.25% crit) to 20% (with 5% crit), so the number of brigands required drops by a factor of five.

Alternatively, does "fly forward to Fire Breath as many of them as possible" mean that the dragon flies at 90ft altitude until it is 45ft past the edge of the army, and then it breathes downwards? If so, then the brigands get an extra turn to attack it, because the dragon can't fly 135ft without taking a dash action.

---

I was not able to find any CR0 creatures that had a viable ranged attack. The CR0 "magewright" has a dagger throw with a maximum range of 60ft, which cannot reach a dragon that is 90ft in the air.

I was not able to find any CR0 creatures with a fly speed over 60ft. To surround the dragon, a creature would need to fly (90ft start range) + (20ft length of dragon) + (15ft wing attack radius) + 5ft = 130ft at minimum. If creatures do not surround the dragon, then it can just wing-attack and leave without taking damage. I don't see a path to a solution involving CR0 creatures flying up to the dragon.

It's surely possible to decrease the number of CR 1/8 creatures that are needed, but I don't see a possibility for a CR0 solution at present.

Answer 2

1 trillion awakened shrubs (CR 0)

Awakened shrubs are indistinguishable from normal shrubs when not moving.

The Dragon cannot fly forever. It sees no enemies. When it lands, it is attacked by shrubs from all sides.

The Dragon is given no opportunity to rest, no opportunity to heal.

After running out of exhaustion levels, it dies.

The Dragon can be induced to move wherever the shrubs want by having a shrub move; it then becomes an enemy, and the dragon is forced to try to move into position to use its fire breath on it.

Once it stops, it is indistinguishable from a normal shrub, and no longer an enemy.

There is no peace, only shrubs.

Answer 3

228 CR 1/2 Scouts (none die)

Scouts are a CR 1/2 from the core rules. They attack the with a longbow at +4 to hit, 1d8+2 piercing (6.5 average), range 150/600 feet. They can shoot the dragon before he ever gets close enough to breathe. (The can also shoot it all the time if it cycles at 350 feet, but lets ignore that here).

The dragon descends from 350 feet. It dashes for at total move of 160, down to 190 feet. The archers ready actions in their turn. The dragon as a legendary action moves another 40 feet, down to 150. The archers shoot their readied actions in response. Then it is their turn, and they shoot again.

The dragon has AC 22, so the to-hit chance with +4 is an 18 or better, 15%. 5% of all attacks are criticals, dealing another d8 or 4.5 damage. The total expected damage per shot is thus 15% x 6.5 + 5% x 4.5, or 1.2 damage per shot.

The dragon has 546 hits, so it needs 546/1.2 or 455 hits. As each scout gets to shoot twice, that needs 228 scouts.

The scouts would need about 15x15 five foot spaces to stand in, or a square field of about 75 feet. At 150 feet range, that is still small enough to have enough range to the dragon to catch him with more than 90 feet distance, even when shooting diagonally.

---

If scouts are allowed to die, 47 CR 1/2

It is possible to do it with fewer scouts if they are allowed to die and need not kill the dragon during the first dive. The dragon will incinerate part of them, and the rest gets to shoot again on the second dive, and so on.

Positioning: For this, the scouts should aim to spread as far apart as they can while still able to hit the dragon, to minimize the impact of the breath weapon. Using a bit of trigonometry, if the dragon descends to a vertical height of 90 feet, the furthest someone can be off on the ground to still reach him with 150 feet range is 120 feet. The dragon might descent at the side instead of in the middle, but it must at least come down directly above scouts so the breath can reach them. We'll assume the scouts spread out in a 120 foot diameter circle. Because the dragon can cover a large share of that circle, better than spreading out evenly is to make four dense cloverleaf clusters at the four points of the compass, so that the dragon at best can blast one quarter of the scouts each time. The scouts can also move during their turn before the dragon breathes, to add another 30 feet and make sure there will be no overlap over groups.

We'll also let them reposition themselves after each dive. They have 30 speed and can dash, plenty to spread out and recover the 90 feet cleared by the dragon while the breath weapon recharges.

Movement Once the dragon got shot twice, it gets to move again: it will move down 40 feet as a legendary action with Wing Attack, descend 20 more feet to 90 feet, and breathe fire. As that is an action, it cannot dash and will only have 60 more feet of speed to move up to 150 feet again. It gets hit by another volley, before it can escape short range.

Death Toll: A 90-foot cone has a diameter of 90 feet at the end, and everyone in the cone would die as the damage is sufficient to kill them, if they make the save or not. The cone will catch one of the four cloverleaf groups that dies, 75% survive. Each dive, the dragon takes damage based on the remaining number. There is some granularity, because of course, a scout can only fully survive or die.

Damage output: As calculated above, on average each scout deals 2.4 damage to the dragon on the way in, and each surviving one deals another 1.2 damage on the way out.

Tabulating this out in a spreadsheet, the minimum number of scouts that can deal a total of 546 and still have someone standing in the end is 47 scouts. 3 of them survive.

Answer 4

Less than 170 Aarakocra Simulacra (CR 1/8)

These creatures are from Storm King's Thunder (See How many simulacra can Sansuri's Simulacrum create? - apparently unlimited, given enough material components and days / spell slots.)

They have some javelins (30 / 120 ft range), but only +4 to hit vs. Harengon Brigands having +5, so their damage/round against AC22 is significantly worse than DanB's answer. They have a fly speed of 50. They can hopefully expose themselves to Frightful Presence (120 ft) with only a small amount getting fried, then let it wear off so they're immune for the day. Then, with readied movement + move on their turn (as in David's pegasus answer) to within 30 ft for attacks without disadvantage. IDK if they can then scatter enough for the dragon to not be able to move away and catch them all with one breath cone, though. If not all, most.

More importantly, like a normal CR1/4 Aarakocra, five of them can summon a (CR5) air elemental:

Summon Air Elemental. Five aarakocra within 30 feet of each other can magically summon an air elemental. Each of the five must use its action and movement on three consecutive turns to perform an aerial dance and must maintain concentration while doing so (as if concentrating on a spell). When all five have finished their third turn of the dance, the elemental appears in an unoccupied space within 60 feet of them. It is friendly toward them and obeys their spoken commands. It remains for 1 hour, until it or all its summoners die, or until any of its summoners dismisses it as a bonus action. A summoner can't perform the dance again until it finishes a short rest. When the elemental returns to the Elemental Plane of Air, any aarakocra within 5 feet of it can return with it.

Air elemental tactics

This is the standard CR5 creature, fly speed 90 ft, multiattack of 2x Slam +8 to hit for 2d8+5 damage. (Against AC22 without disadvantage, one Slam averages 6/20 * (2*4.5 + 5) + 1/20 * (4*4.5 + 5) = 5.35 damage, for 10.7 DPR for a multiattack.) If they're frightened, they can still Whirlwind for 3d8+2 damage (DC13 Str for half), so 7.8 average damage (saving on 3 or higher.)

10.7 damage per multiattack means ~51.03 elementals will take down an ancient red in the first round. 255 Aarakocra can summon 51 elementals, and do enough extra damage on their own to finish off whatever HP the dragon last est. Depending on tactics, in the worst case (the elementals end their turn next to the dragon and its breath recharges right away) all the elementals get one more Slam as an opportunity attack before the dragon moves and fries them all, with only 10% making their save to survive the breath. In which case 34.02 elementals, or only 170 Aarakocra, on average almost killing it with 3 slams per elemental. I think any other tactical developments will result in more average attacks per elemental before going down, such as elementals scattering again, but I'm not sure exactly how it would play out.

They're faster than the dragon by 10 ft, but without the legendary action for extra movement. Total speed per round in initiative (not chase rules) = 2x90 = 180 ft, vs. the dragon with 2.5x 80 = 200. Air elementals are immune to exhaustion, dragons aren't. So I'm going to assume they can get more or less whatever starting position they want, such as surrounding the dragon on all sides, so a breath couldn't hit many of them.

(I haven't thought about how best to coordinate this with the aarakocra, whether it's worth sacrificing any elemental tactics to have the aarakocra be able to swarm in and attack in the same round after it uses its breath weapon. After they finish summoning, it's not concentration and the elemental is only lost if all 5 summoners die, so teams should spread out to a breath weapon can't hit all five. Or have one aarakocra from each group stay far back.)

Its average HP total is 90, just 1 short of the average 91 damage of a fire breath. (Its dex save is +5, so only 10% of them make the DC24 save for half and survive). With readied movement, they're fast enough to surround the dragon. They have resistance to bludgeoning, piercing, and slashing from nonmagical attacks, so it helps against any melee attacks but not against Wing Attack (which is mechanically a save, not an attack roll.) That seems dumb (especially with "attack" in the name), but RAW is RAW. They are immune to Prone, so at least they don't fall out of the sky or even have disadvantage on op attacks if the dragon uses the movement part of wing attack.

If most of them started 100 ft away, they could have 80 ft of movement left after attacking, to fan back out into a big cloud. (Otherwise the dragon moves away with a wing attack before its turn, taking an opportunity attack from each of them, but can then hit them all with a breath on its turn because they're all grouped up where the dragon was. Ones that save will still survive, but the wing attack damage makes it very unlikely for a low damage roll to let any survive that don't make their save, or a negligible number if you randomize their HP as 12d10+24 instead of all having 90.) If they gamble on the breath weapon not recharging on the first turn and stay in, the dragon can wing attack + dash to move 200 ft away even without its breath weapon, so only a few of the elementals could get within melee range with their own dash, to be able to make op attacks if the dragon keeps fleeing.

From opposite sides of the 20-foot cube occupied by the dragon, they could get back out to a sphere about 2x 80 + 20 = 200 feet diameter, using their remaining 80 feet of movement to go away from the dragon after initially surrounding it on all sides to make melee attacks. (Or 210 since they can be 5 ft away while attacking. Air elementals could enter the dragon's space without squeezing, but they wouldn't want to). A 90 foot cone AoE is 90 feet across at its widest point, but narrower before that, so it only hits a fairly limited number, probably well under half.

Rinse, repeat with surviving elementals. (One of them takes an attack of opportunity on the way out, which only misses on a nat1, using the dragon's 1 reaction. They're resistant to the physical part, but not the fire from a bite.)

The dragon doesn't have enough speed advantage to disengage and get far enough away that the elementals can't catch up by dashing themselves, even with a wing attack after its turn before the elementals' turn.

Elementals aren't immune to the frightened condition, so they might want to try to trigger Frightful Presence and retreat until it clears before coming in again, if they plan to stay near the dragon while it takes a turn. This will result in some losses, but I expect less than 50% so it's better than having them use Help actions. But the dragon can only use Frightful presence on its own turn, and only as part of its multiattack, not when using its breath weapon. So it won't come into play on the first turn the elementals move in to melee, only if they stay in and gamble on the breath weapon not recharging.

---

Honourable mentions: CR 1/4 flyers immune to fire:

• Metallic Warbler from Fizban's, 14 HP, fly 60, one melee attack at +4 for 4 damage (1d4+2).
• Ashen Flying Sword from Keys to the Golden Vault, 17 HP, fly 50, one melee attack at +3 for 5 damage (1d8+1).

They're fast enough to swarm in with a readied action plus movement on their turn, as in David's pegasus answer. But they don't have to be clever about it because they're immune to fire damage. However, a Wing Attack AoE will destroy them all with an average roll (17 damage from 2d6+10).

Answer 5

Here is an obvious answer (just catch up to the dragon) to set a baseline.

323 CR 2s

A flerd (a portmanteau of flock and herd as suggested by TimothyAWiseman in the comments) of Pegasuses can close the gap and slowly whittle down the dragon.

The Pegasuses have enough flying speed for one Pegasus to make it to the dragon after it uses its Fire Breath. Here is the tactic the flerd employs:

• A single Pegasus gets within 90 feet when the dragon allows it while the rest stay out of Fire Breath range readying their movement to have one run forward (only one so they won't all get caught up in a lucky Fire Breath recharge) until after the Fire Breath
• After the dragon uses Fire Breath, the one Pegasus's Readied action triggers allowing it to get closer. Then on its turn it can close the rest of the gap and make a single Hooves attack.
• The dragon then uses Wing Attack making the Pegasus fall prone and possibly dying to the combination of Wing Attack and fall damage, and then resumes its kiting.¹

Every instance of this tactic, the flerd has a chance to lose a Pegasus to Fire Breath, or to Wing Attack. If either survive, they are chosen for the next cycle. This means the reduction of the flerd per cycle is as follows:

Fire Breath deaths

• 95% chance that the Pegasus fails the Fire Breath save
• On a failed save, there is a 0.01% chance that the dragon doesn't deal the 59 damage required.
• 5% chance that the Pegasus succeeds on the Fire Breath save (thanks to it's +4 Dexterity save)
• On a successful save, there is a 99.9% chance that the dragon doesn't deal the 118 damage required.

Overall, there is a 5.95% chance that the Pegasus survives for a second cycle; as such, let's find the 2-cycle death count and divide by 2:

• At the start of a 2-cycle we either have a full health Pegasus or a damaged Pegasus.
• In the full health case (94.40% of the time) we get the following outcomes
  • There is an 88.46% chance that both Pegasuses over two cycles die for 2 deaths
  • There is a 5.95% chance that the first Pegasus succeeds which then dies in the next cycle for 1 death.
  • There is a 5.59% chance that the first Pegasus fails, but the second one succeeds for 1 death.
• In the damaged case (5.59% of the time), the first cycle results in a guaranteed Pegasus death (the chance it still lives is negligible), and the second cycle uses the normal 5.95% chance of survival.

Overall, this yields a 2-cycle death count from Fire Breath of 1.89, or a 1-cycle death average of 0.94.

Wing Attack deaths

• 100% chance that the Pegasus fails the Wing Attack save
• The total of the Wing Attack and the fall damage is 12d6 + 10. Just like before, this is guaranteed to kill the Pegasus on a 2 cycle.
• In the full health case (88.13% of the time) we get the following outcomes
  • There is an 74.37% chance that both Pegasuses over two cycles die for 2 deaths
  • There is a 13.76% chance that the first Pegasus succeeds which then dies in the next cycle for 1 death.
  • There is a 11.87% chance that the first Pegasus fails, but the second one succeeds for 1 death.
• In the damaged case (11.87% of the time), the first cycle results in a guaranteed Pegasus death (the chance it still lives is negligible), and the second cycle uses the normal 13.76% chance of survival.

Overall, this yields a 2-cycle death count from Wing Attack of 1.76, or a 1-cycle death average of 0.88.

In total, the flerd loses 1.82 Pegasuses per cycle. Therefore, we just need to calculate how many cycles it will take to kill the dragon to reach our flerd size.

Hooves Damage

Our Pegasus has a +6 to hit which actually can hit the dragon 25% of the time. Here is the damage it can deal:

• An average of 11 damage on a normal hit.
• A 5% chance of dealing a critical hit for an extra 7 damage.

This totals an average of 3.1 damage per cycle.

As such, the Pegasuses kill the dragon after 176.13 cycles on average, losing 320.55 members on average. Therefore (rounding up the death count for safety), since 2 are needed to trigger the cycle, a flerd of 323 Pegasuses, can kill the Ancient Red Dragon on average.

^{1: You may say the dragon should kill this Pegasus on it's turn rather than fly away, but using this tactic would allow the other Pegasuses to close the gap as well since the dragon isn't able to kill the Pegasus consistently with Legendary Actions alone.}

Answer 6

One CR2 Young Fire Giant

The Young Fire Giant from Tales from the Yawning Portal has a javelin attack with range 30/120, and is immune to fire.

In combat, the red dragon uses its fire breath on the giant, over and over and over, from 90ft away. This does zero damage. In exchange, the giant throws javelins from long range, over and over and over, and occasionally hits.

The combat tactics you specified are focused on kiting, and the dragon never realizes that it could charge into melee.

The javelin is a +6 attack with disadvantage vs AC 22, so the chance to hit is 1/16. If we neglect critical hits, then each hit does 11 damage. We can estimate that the dragon dies after (546hp) / (11 damage) / (1/16 chance to hit) * (3 turns per fire breath recharge) = 2382 rounds.

Rules As Written, the giant does not run out of javelins. Edit: yes it does, apparently -- "A monster carries enough ammunition to make its ranged attacks. You can assume that a monster has 2d4 pieces of ammunition for a thrown weapon attack, and 2d10 pieces of ammunition for a projectile weapon such as a bow or crossbow."

Answer 7

62 CR 1/4 Goblins

Goblins are my fav minion, so let's see how well they fare. The goblins will employ standard gobbo tactics; attack at night, hide, spread out, and run as soon as they get spotted.

Goblins have +6 Stealth, they are in darkness so they had advantage on their check. The dragon has 26 passive perception, but is trying to use Darkvision in darkness so they are at disadvantage, giving them 21 passive perception. This gives a goblin a 43.75% chance of hiding. Unfortunately thanks to Blindsight not having any penalties from darkness, goblins will be unable to hide within 60ft of the dragon. The goblins will continually attempt to hide if they are detected - they can hide twice per turn if they fail, giving them a 68.36% chance to hide if they don't attack.

The goblins within 170ft of the dragon will position themselves far enough apart to prevent more than one being hit by the dragon's breath attack. The rest of the goblins will prefer to be 170ft away from the dragon. Assuming the dragon recharges firebreath in 3 turns on average, on the first turn it will move 210 feet away on its first turn, which means roughly a third of the goblins will be out of range - their short bows have 80/320 feet range. On the second turn it moves another 210 feet, at which point all of the goblins will be out of range. The goblin will then spread out once again to be far enough away to prevent more than one being hit by a firebreath.

Goblins have +4 hit vs the dragon's 22 AC, they are attacking a target at long range (disadvantage) against a target who can't see them (advantage), so they make straight rolls. They have a 15% chance to hit, so considering their 1d6 + 2 shortbows they deal an average of 1 per attack.

That means the goblins need 546 attacks in total. For every 1 goblin death we get 1 round of full attacks, and 1 round of 2/3rd attacks. Each goblin has 11 arrows on average, so we are limited to 11 rounds total. Brute forcing it, I can see that 62 goblins is enough.

1. dragon spots a goblin and attacks, 61 goblins left. Goblins attack, dragon has 485 hp
2. dragon flies away, 40.6 gobs attack (on average), dragon has 445 hp
3. dragon is out of range, regenerates firebreath
4. dragon spots a goblin and attacks, 60 goblins left ...

You get the idea. It's hard to model the exact behaviors, but this should give a rough idea. Any other monster with a long range attack and fast movement speed should be able to spread out enough to prevent the dragon being effective while also being able to attack.

The dragon can increase its chances by using its legendary Detect action to search for gobs (that said, it does require DM ruling as to whether you can detect multiple goblins, if you need to specify the area you are searching, or even if this is equivalent to the search action and its interactions with hiding, etc), then try to stay on the outside of the goblins. This would increase the number of goblins needed, because less of them will be able to attack the dragon at any given time, and they may be unable to hide properly which would force them to attack at disadvantage.

Answer 8

21-41 eagles (CR0)

and 20-40 legacy¹ swarms of rot grubs (CR 1/2)

We start with 21 eagles (Size Small, Fly speed 60') and 20 swarms of rot grubs (Size Tiny, 5' climb). We could use hawks instead of eagles, since they have the same relevant stats.

The rot grubs mount the eagles, at one swarm per eagle (the eagles must be a size larger than the swarm, which they are, and must be of appropriate anatomy, which is a DM's call).

The eagles disperse and ascend to 350 feet, the cruising altitude of the dragon, to keep the math simple. As the dragon approaches them to breathe, they keep one sacrificial eagle 30 feet in front of the rest, as in the flerd of Pegasi example, but are maneuvering, dashing at 120' as necessary to insure that the dragon can only breathe after it has taken its full movement, or can retreat only 10 feet.

The dragon breathes on the sacrificial eagle at 90 feet and ends its turn.

Twenty more eagles dash up to 120 feet and enter the space of the dragon. They can enter the dragon's space because it is at least two sizes larger than they are (it is, in fact, four sizes larger than the eagles and five sizes larger than the swarms). The swarms have blindsight to 10' - as soon as the eagles enter the dragon's space the swarms use reactions to take their readied actions to attack (their attacks have a reach of zero, so they must be in the dragon's space to attack).

With 20 swarms of rot grubs, on average one will score a hit on the dragon with a 20N (they have a +0 to hit, so need a base 22 to hit the dragon's AC and hit only on a 20N). On a hit, the dragon is infested by 1d4 rot grubs (average 2.5).

At the end of the eagles' turn the dragon uses its wing attack. Neither the eagles nor the swarms of rot grubs can make a DC25 Dex save; all of them are killed. The d4 individual rot grubs that have successfully infested the dragon survive.

At the start of the dragon's next turn, it takes d6 piercing damage per rot grub (average 8.75 damage). If it can recharge its fire breath, it can breathe on the wound during its turn and will kill the rot grubs; but there is a 67% chance that it cannot and the 20 swarms and 21 eagles will have been enough. Or, the DM could rule that since the dragon is immune to fire damage, it cannot do damage to itself and its wound and therefore the rot grubs are safe.

If a rot grub in a wound is not safe from the dragon's fire, and we want to make sure we succeed even if the dragon's breath recharges, we can use 41 eagles and 40 swarms of rot grubs to make sure that there are at least two wounds, and then spread the eagles out over all sides of the dragon's body so that it is unlikely that the dragon can breathe on both wounds with the same breath.

After the end of the dragon's first turn after it was infested, the rot grubs will be safe from further attacks. They will continue to burrow through the dragon until they reach its heart and kill it. This will take on average about ten minutes.

---

¹ I own VGM but not MPMM. As pointed out by smbailey in comments, swarms of rot grubs were updated to do poison rather than piercing damage, with the damage ending upon a single Con save rather than the death of their host. While the VGM grubs are truer to the original vision for the monster, I can see how a CR1/2 monster that can kill an ancient red dragon with a single hit could be problematic for 'balance'.