Without question, the unofficial motto of the Slowtwitch forums is, "Your seat's too high." While this has become a bit of an inside joke, there's a fundamental truth in this quip that's worth discussing further. The best height for one's saddle is the right height. But the second best seat height is one that's too low. Rightness of saddle height does not exist on a bell curve. A saddle that is only slightly too tall is substantially more problematic than a saddle that is even a great deal too low. And the research bears this out.
Even with a saddle height that is overwhelmingly low, performance does not degrade a great deal. You'll generate less power, and it might be slightly worse for triathletes because a too-low saddle height will also end up closing off your hip angle, but what you won't do is increase your risk of injury. A saddle that is even slightly too tall can lead to all kinds of problems from the annoying-but-minor, like saddle sores from shifting left-to-right to prevent reaching, up to more significant issues in the SI-joint, the hamstrings and calves, and more.
This is why the right way to determine your proper saddle height is to start with a height that is very obviously much too low (say 0.70x inseam length; no one's saddle is this low) and to incrementally raise it until you are happy. If you start with a saddle height that is too high - especially quite a bit too high - the sense of relief from lowering it can cloud your judgement about what's appropriate. If you start with a saddle that's too high, you may lower it but still end up with a saddle that's too high, and that's a problem. If you start with a saddle that's too low, you may raise it but still end up with a saddle that's too low, but that's okay.
This is especially true because the tallest you will ever want your saddle is probably the height that you pick when you are doing a bike fit. 100 miles into an Ironman bike ride, when you are tired, and sore, and dehydrated, and ready to get off your bike, that saddle height is going to seem much too low when you are fresh and pedaling in the comfort of a fit studio. It's not uncommon for Grand Tour cyclists to finish a three-week race with a saddle height that is 10mm - or more - lower than what they started the race with due to fatigue and muscle damage. This is another reason why having a saddle height that, during your fit or when you first start spinning, seems to low is not only not a problem, it's actually advisable.
But this is Slowtwitch, where, in addition to everyone's seat being too high, our second unofficial motto is, "In The Lavender Room We Trust, all others bring data." Really, we don't trust anyone, and everyone needs to bring data, but that doesn't sound quite as good. So let's look at what the research says here.
The first study is, "Acute Effects of Small Changes in Bicycle Saddle Height on Gross Efficiency and Lower Limb Kinematics" by V. Ferrer-Roca et al [full text available]. This study did not focus on what is a "correct" saddle height; it operates under the assumption that, in the case of well-trained cyclists (n=14), saddle height is necessarily appropriate. Which makes sense. I've never been a fan of saddle-height-as-a-function-of-trochanter-height or any of the other static metrics. They simply don't account for variances in either pedaling dynamics (toe-pointer versus heel-dropper) or in simple - and too often overlooked - personal preference. The study conducted three randomized trials of 6min at a constant (though not self-selected) cadence of 90rpm and an intensity of approximately 70-75% VO2Max. Saddle height was preferred saddle height, 2% higher, and 2% lower. For more people, a 2% change in saddle height is dramatic. In my case - saddle height of 823mm - that's over 16mm, which is enormous. A saddle that is 2% low is really low. Likewise, a saddle that is 2% high is really high. The findings?
Gross efficiency was significantly lower and oxygen consumption (v_O2) was significantly higher when raising the saddle (GE = 19.9 6 1.5%; V_O2max = 43.8 ml/kg/min) than when lowering it (GE = 20.4 6 1.3%; V_O2=42.8 ml/kg/min).
GE (%): 20.4±1.3 -- 20.3±1.8 -- 19.9±1.5
VO2: 42.8±4.9 -- 43.3±4.9 -- 43.8±4.9
You can see here that gross efficiency is virtually unchanged when the saddle is lowered, even by an extreme amount. In fact, the highest GE and the lowest VO2 was reported with the lowest saddle height, indicating that even for these cyclists, Slowtwitch knows best. Their saddles are too high.
Another interesting finding was that, as Dan and I have repeatedly seen during fit workshops, the body "protects" the knee and hip angle by increasing the ROM of the ankle. In all trials, hip angle deviated by less than 1° (on average) from preferred; knee angle deviated by about 1°; but ankle angle deviated by about 2°. This is especially true for a saddle that's too high. In this case, the body just really seeks to prevent overextending at the knee and hip. For bike fitters, this means that focusing on the ankle can be the most reliable way to work with a client to help zero in on the correct seat height; someone who "reaches" with the foot is almost certainly sitting with too tall a saddle.
Hip Angle ROM: 45.8±3.1 -- 46.0±3.2 -- 46.±2.9
(2% low, preferred, 2% high)
Knee Angle ROM: 74.1±5.6 -- 75.5±5.1 -- 77.3±4.3
Ankle Angle ROM: 10.6±4.7 -- 12.3±4.2 -- 14.8±7.2
This finding is supported by a larger analysis of the research to date, presented by B. Sather & D. Dutto in their presentation, "Effect of Seat Height on Cycling Efficiency." This gives a nice review of many other studies on the subject for those who want to dive even deeper. Sather & Dutto mention the findings of Peveler's 2008 study [see Addendum below], "Effects of saddle height on economy in cycling" [PubMed]:
VO2 was significantly lower at 25° knee angle compared to both the 35° knee angle and the Hamley method of 109% of inseam.
And Price and Donne's 1997 study, "Effect of variation in seat tube angle at different seat heights on submaximal cycling performance in man." [PubMed]:
104% of trochanteric height (knee angle 157.5°), VO2 and heart rate significantly higher and power efficiency significantly lower than both 96 (146.4°) and 100% (knee angle 136.9°)
Taken together, the science is pretty clear. Your seat actually is too high... Yet again, the wisdom of crowds - at least a crowd of STers - proves it's worth.
[Addendum 2017.02.27] As was rightly pointed out in the comments below, Peveler's study actually showed an decrease in VO2 with a taller saddle height. I mixed this up because I'm used to the FIST standard of using the included knee angle, where a small number is a more bent leg; the use of the external knee angle still throws me sometimes, because I find it illogical. That said, I did mix this up. The conflicting results here are addressed in Stefanov and Kolev's 2016 study, "Cyclists Saddle Height Importance on the Results of Bicycle Ergometer Test - 'Astrand and Rhyming.'" This looked at 70% and 100% saddle height and found the same conclusion as Ferrer-Roca et al. Lower saddle height resulted in lower VO2. But Stefanov and Kolev explain this as an artifact as to how VO2 is calculated. They did not test a taller saddle height, but they support the idea that a very low - 70% is way too low - saddle is not actually much of a problem.