I know the smooth ground and no wind resistance make it easier. But does the ground being pulled beneath you make it easier too?
The science seems divided, though I am inclined to see value in the arguments that differences persist.
Update the first: Though the physics "frame of reference" solution has merit, I am for the moment convinced by the idea that the addition of energy from the belt-rotating mechanism changes the system significantly enough for treadmill running to differ from overground running kinematically, bio-mechanically, proprioceptively, and in terms of inertial reference frames. When I'm running overground, my foot lands on a surface that is moving in reference to my body, but it will not continue to move at that speed unless I impart force to it. This is not true on a treadmill. We are adding outside forces and momentums with the addition of a belt instead of a massive earth.
Moving Through Space Versus Swinging The Leg
This RunnersWorld article has a high-level review of some literature on differences between overland and treadmill biomechanics.
Bryan Chung of Evidence-Based Fitness made a nice argument against the idea that treadmill running is similar to ground traversal running. Unfortunately, as I recall, there were some clarifying points in the comments that are no longer visible since his switch to Facebook comments.
There's also, as Chung indicates, a good deal of biomechanical analysis of the topic.
Lee and Hidler argue that the treadmill is fine for therapeutic uses, but found muscular differences. Elliot, Pyke, Roberts, and Morton (PDF), contrary to Chung's argument, found no differences in gait. However, they did not make any analysis regarding energy expenditure and are careful to note that their research was solely on sub-maximal running mechanics. Parvataneni and Krishnaji found that the elderly (including stroke patients) had a greater energy expenditure when on a treadmill. They also found, as others have, that overground walking involves much more stabilization work necessary to combat asymmetry. Alton, Baldey, Caplan, Morrissey found significant differences in gait attributes. (Their research was restricted to walking.)
I would expect that the disagreements over whether differences exist is due to different definitions of gait mechanics, and/or the subjects used. I am therefore very tentative in any conclusions I draw from these studies.
Update the second: Doctor Chung has kindly updated his post with a set of studies from the old comments:
I particularly appreciated the methodology (alas, gleaned from the abstract alone) in this study by Frishberg:
I am also immediately struck by the inherent problem in comparing overground sprints, where maximal speed is determined by the runner, with treadmill sprints, where speed is dictated by the machine controlling the belt.
In contrast, this study by Bassett, David R. Jr., Giese, Nagle, Ward, Raab, and Balke tried to study the same thing, and I think its methodology is deeply flawed. There's also this study on walking, which shows exactly what one would expect if people were allowing the belt to move their feet for them:
That sounds to me (though I could be wrong) like late in each step, the foot is allowed to come up instead of continuing to push against the ground.
The MovNat-Style Perspective
What the heck are you doing inside paying money to look at an LCD readout when you could be outside--for free!--seeing beautiful places and climbing rocks and getting tired the way humans have gotten themselves tired for millennia! Gah! "Don't take a perfectly good activity like walking and make it suck by doing it on a treadmill." (quoted from Your Workout Sucks on YouTube)
Unless you have a specific reason to use a treadmill--you're recovering from an injury, you live in Antarctica, you're training for the Inside Olympics--the kinematics and biomechanics are probably insignificant compared to the benefits of running outside. More importantly, you should find it aesthetically offensive. For example, strength and conditioning coach Keegan Smith has this response to someone with a similar question. It loses nothing by substituting "treadmill" for "elliptical":
No. The belt "being pulled beneath the runner" is not a factor. As stated in previous answers and comments, there is no privileged frame of reference, so it is just as valid to analyze the treadmill system based on a coordinate system that moves with the belt as it is to analyze the overground running based on a coordinate system that is fixed to the Earth. These two systems are mechanically equivalent (if the ground happened to be the same material as the treadmill). This was established early on (1980) in the literature, and hasn't been challenged since.
There is another factor to consider about the ease of use of running on a treadmill versus running outside, how it affects you mentally. I find it much more difficult to run on a treadmill because I find it incredibly monotonous compared with the variety of sensory input experienced while running outside. The MovNat reference that was mentioned extends that even further. Keep in mind, too, that even slight terrain variations within a city can help provide a bit more physical variation (texture, slope, camber, etc. and the resulting slight muscular compensatory differences) compared with the exact same motion on each step on a treadmill. There is an additional physical difference in your ability to change motion as you feel the need to without needing to manipulate an external control first. Unless the weather makes running outside treacherous, I would much rather do that than run on a treadmill.
Edit: See the other answers for why I have changed my position. disregard the following.
I disagree with the other answer. The belt pulls your foot from the moment of impact until you lift your foot again, reducing the work that your hamstrings do.
One recommendation is to adjust the incline on the treadmill to 4% to adjust for the disparity.
In conclusion, yes.
Like most things, yes and no, and some of it is individual. Personally, I find running outside to be much more physically challenging. Part of it is that, as your question notes, the treadmill does create a "pulling" effect, though I suspect that the majority of the difference is the basic and often minor variation of terrain that your muscles constantly adjust to. My experience is that I can hit impossibly fast mile times (for my own fitness level) on treadmills than outdoors. On the other hand, though, treadmills don't just pull your feet; they also force you to run faster than you would generally push yourself on your own.
Treadmills are easier, but it won't be a big difference (unless it is raining, freezing, or very hot outside), which is probably what friz meant.
Someone on a treadmill could easily cheat by cranking the speed way up and hanging onto the rails, letting the motor pull their legs. But then that is cheating. You can't really cheat when you are walking on the ground.
You are not physically moving the mass of your body, but the motion of moving your legs and arms are still going to cause your heart rate to increase and calories to burn.
Walking "in the wild" (i.e. on the ground) can burn more calories because of the extra work required to move the mass ...but it should be negligible in the grand scheme of things given that treadmills are so versatile.