According to what I've read, SpaceX tried to use parachutes to recover the first stage of the Falcon 9, but it did not survive reentry.
Now they plan to use 2 separate rocket burns to land:
I understand the reentry burn in light of the previous failure, but my question is why not still use parachutes instead of a landing burn?
There are many key points to this, probably none on their own sufficient to ditch the parachutes approach (except economics, those are good enough on their own), but together they make for a compelling case against it;
Descent control: As already mentioned, there's a significant guidance uncertainty with the use of a parachute system. Some of it comes from the weather in the lower atmosphere, and some from the geometry of booster stage and where the forces during descent are applied to it. Weather uncertainties mean that you'll have a wide and long best effort landing ellipse where the stage could be retrieved, and with slightly unfavorable weather downrange, further limits your launch opportunities. This might be excusable for NASA's human spaceflight endeavors (I personally don't share that opinion, STS could barely launch at all due to all the safety constraints imposed on it, and such parameters don't seem to stop Russians so much from launching). The other problem is that you can't correct for any such drift as easily with propulsive system due to stage geometry.
Structural stability: Geometry of the stage, being a long cylinder with its center of gravity low and close to the engines with only a certain percentage of propellants remaining, means that parachutes will apply force on the stage at the opposite end of it while its descent rate is reduced. This is not optimal. Deploying parachutes is also rather violent in terms of achievable maximum g-forces even during staged parachute deployment. This could introduce significant lateral forces to the stage if the parachutes deployed while the stage's long axis isn't aligned with its velocity vector. Rocket stages are not constructed to withstand much lateral force so they can remain light, and most forces associated with their launch and during staging are along their long axis, not lateral.
System complexity and reliability: Adding an additional, complex system of pilot, drogue and main chutes adds to the overall complexity of the already complex system, increasing chances of something going wrong. Reliability of parachute descent systems and their final design and reefing profile, is also hard to establish without first sacrificing a great deal worth of flight hardware. This is a rather tentative process with many uncertainties, and unless you have long years of testing and development behind it (like, say, NASA's Orion that has nearly 60 years long legacy in Apollo program), and can design your vehicle to withstand structural loads, it's not very likely that you'll get it right for an entirely different system before your budget runs out.
Weight (argumentative): Depending on parachutes wouldn't save much weight, if any. It's a complex system with many components, including (but not limited to) parachute fairing that would protect it during ascent, staging and hypersonic part of the descent, cabling for pilot, rogue and main chutes and a system to cut the lines when needed, canopies themselves and many additional sensors and actuators needed to make it happen. The stage would also require structural reinforcements to provide parachute attach points and withstand forces they'd introduce to the descent profile. Using parachutes also doesn't get rid of de-orbit, de-spin / stage rotation, and final landing cushioning retrofires. Such system might also require additional inflatable airbags to cushion forces associated with landing. Keeping some 10% of propellants in your tanks for the final landing burn and slowing down from stage's terminal velocity to near zero vertical speed seems pretty close to the added weight a parachute descent system would introduce.
Manufacturing volume: SpaceX is keen on lowering price of launching to space wherever possible. Serializing manufacturing processes while keeping them flexible and easy to implement future improvements (often by introducing innovative manufacturing techniques such as additive manufacturing, aka 3D printing) is key to keeping non-operational costs down. So while they offer to their customers an option - fly on expendable, more capable hardware and pay more, or fly on reusable, slightly less capable hardware and pay less - they do this, or rather - will do, with exactly the same parts as far as manufacturing is concerned. There are but minor differences between F9E and F9R, most of them modular, such as addition of the grid fins and landing legs to otherwise same, or nearly same stage. Using parachutes on F9R and no parachutes on F9E would go against such doctrine and require two different stages, one reinforced to support parachutes, and an expendable, lighter version that wouldn't require that. So you'd have two manufacturing lines instead of one. Not economical.
And there are some other, finer points that I'll omit to keep the length of this answer down. But I believe we can agree that going for a system that's already an essential part of the stage was the better option.
There's another way of looking at it, tho. Reusable hardware has one inherent property that it is worthless unless proven to still work. So, if you fail at recovering the stage due to, say, descent engine failure, and there's not enough time to compensate with the other of the nine Merlin 1D engines because they also didn't fire in time, then most of the value of your hardware you were trying to recover is near worthless anyway. How exactly will SpaceX be discarding reusable hardware once it's tested to no longer be suitable for reuse is still somewhat a mystery, but failure to recover the stage would be obviously one of the ways to do that.
For more information also see a related question What was SpaceX's original vision for booster reusability? and I'd also recommend Coming Home: Reentry and Recovery from Space (free download in EPUB / MOBI / PDF formats via NASA e-Books page) as a general reference book.
Because the F9 rocket's first stage already has nine rocket engines.
Why isn't a helicopter landed with parachutes? Because it has an engine and rotating wings. It can use them to start, fly and land. Same thing with rocket engines.
The Falcon 9 first stage uses 3 engine burns in its landing trajectory:
Boostback, to kill its forward velocity and return in the general direction of the landing pad. This decelerates the stage from 5000 km/h to 0, plus a bit of return speed.
Reentry, to reduce its speed as it enters the atmosphere. Starts at 45 km up, the stage is supersonic at the start of this burn.
Landing, to steer the stage toward the landing pad and kill off its remaining speed. Note: if the engine fails at this stage, the current trajectory will have the rocket crash into the ocean. So, how much remaining speed are we talking about?
According to the post-landing teleconference, the sonic booms arrived at a point 4 miles from the landing pad at about the same time that the rocket landed. Taking a few shortcuts, that means the booms originated 20 seconds before landing, or well into the landing burn (which started 30 s before landing). So the landing burn starts at a speed of 400-500 m/s.
Now, if you want to replace just the landing burn with a parachute, you'll have to do something about that initial speed. Parachutes that work at supersonic speeds are rare: the only ones I know of have been built for Mars, not Earth. So you need to extend the reentry burn to decelerate the stage to something like 150 m/s.
You'd also need just about the largest parachute ever built to decelerate the 23-ton stage. This parachute weighs about 1 ton.
So instead of a rocket burn with a delta-V of 500 m/s, you get a rocket burn with a delta-V of 350 m/s (1) plus a deployment of the largest parachute ever made, plus another rocket burn to cushion the landing.
This seems to add a lot of complication for little benefit, especially when you add in the fact that you also need to steer that parachute because your existing steering mechanism (rocket nozzle plus grid fins) don't work anymore when you're under a parachute.
You've also inserted more points of failure into the system.
Finally, parachutes add a lot of work to refurbishing the rocket: they have to be inspected, carefully packed etc. And they add complexity to the landing: say the rocket lands on a barge in the ocean. The rocket's upright, but the parachutes float down into the water where they quickly fill up and start pulling the top of the rocket sideways, threatening to tip it over. You'd have to have a quick and foolproof way of disconnecting the parachute from the rocket while not letting it sink to the bottom of the ocean. Remember the barge is unmanned during the landing so this would have to be done automatically.
1: assuming the rocket speed at the end of the reentry burn is 500 m/s
Parachute descents cannot be guided with any accuracy, if you pop the chutes at high altitude the stage would get carried miles away from the landing pad and the stage would not have enough residual fuel for that much lateral movement.
I thought I had heard Musk say that a rocket landing was for another very good reason, over and above those already listed:
Practice/development for landing large mass on the moon, Mars, asteroids, etc.
No atmosphere (or little) means parachutes are simply not an option, especially as the mass being delivered goes up. And the mass will go up a lot with anything past the current slew of 'sensor drones'.
If I have not heard it, I still reckon it is probably true. A vendor with a proven rocket-landing solution is in the box seat for the next generation of space cargo. As if Musk doesn't know that...
EDIT: Said another way, this is a case of 'design re-usability'. Get this design working well on Earth, and it can be taken to other off-Earth destinations with relatively minor adjustments (Mostly scale). - Minor compared with totally changing the landing paradigm.
If you used a drogue parachute that is large enough to slow the vehicle down but small enough not to induce drift you could jettison that before landing under booster power.
No one has mentioned the violence of a hot rocket hitting water at an odd angle. Rockets are delicate. Water is heavy. The force of it jetting trough crevices would probably rip wires and tubes off. There is the danger of it shorting circuits. It would blow steam off the nozzles. It's salty, it's dirty. It would penetrate every part of the rocket that wasn't sealed. It would get into the combustion chamber. What if you dropped your car into the sea from about 15 feet? I think it would need more repair than just drying it off.
Reliable parachute systems for spacecraft recovery have existed and been used successfully for decades, so all of the "parachute systems are too complex and unreliable" comments don't make a lot of sense. It would make it more difficult to land on a platform with parachutes, but why would you need to? A spacecraft landed and floating within a hundred square mile area of ocean can easily be fitted with GPS transponders and located by a vessel equipped to retrieve the boosters. The Apollo spacecraft landed consistently within a few miles of their recovery vessels, so it's very achievable. The comments about requiring a retro burn to slow the craft before parachute deployment also do not track. There are plenty of ways to slow a spacecraft that is reentering the atmosphere that do not require a breaking burn. Atmospheric drag is free and more than sufficient to slow such a craft to safe parachute deployment speed if the craft is designed correctly (again, the technology required to do this is well-developed and reliable). Parachute-assisted ocean landed SRBs for the space shuttle were successfully reused several times over the course of the program. It's been done. It works. The reason Spacex lands their boosters on automated floating platforms is that Elon Musk thinks it's cool. That's it.
