Endurance ExerciseMost of the relevant research I've been able to find focuses on "submaximal exercise" or endurance training. Here the results are inconsistent, and the inconsistencies are themselves interesting to examine. In this study, titled "High fat diets and exercise performance", the author cites animal studies consistently showing significant performance enhancement after adaptation to high-fat diets. He then describes several human trials, some of which showed improved endurance and some of which did not. He proposes an interesting hypothesis to explain the disparity:
Part of the inconsistency between the animal and human results may be explained by differences in the percentage of calories from carbohydrate. The high fat diets given to animals contained little carbohydrate, usually 0-1% of total calories. The human diets on the other hand typically contained a moderate amount (10-20%) of their energy from carbohydrates.
A diet containing 10-20% carbohydrates is not likely to result in ketosis, unless the diet is also calorie-restricted. I assume they were trying not to kill the athletes, since as we all know dietary carbohydrates are absolutely necessary.
I mentioned adaptation. That's a recurring theme in modern research in this area. As most low-carbers are aware, the first few days or weeks of a ketogenic diet are a bit rough, and these scholarly efforts consistently demonstrate this. Many early experiments were too short, either by design or because unexpectedly poor results discouraged researchers or participants. Once the athlete adapts to the diet (enters ketosis and depletes his insulin levels), things turn around quickly and dramatically.
Adaptation is presented here as another potential resolution to the inconsistency of results among human studies, most of which only lasted one to two weeks. This is understandable, since locking participants in a metabolic ward for months at a time to ensure compliance is very expensive.
High-Intensity ExerciseAll the material I've been able to find online suffers from the adaptation problem. One study published in 1994 compensated for this, and found "no difference" in performance between high-fat and high-carbohydrate diets. It's not available online (believe me, I tried) but it's well-summarized here. While it does account for adaptation, even this study provided 7% of calories from carbohydrate, so we don't know whether participants were in ketosis. Given a current lack of available time trial studies that meet the ideal criteria, let's discuss the results our theory would predict, and see if future research can confirm or falsify it.
Performance in high-intensity exercise like sprinting depends mainly on two factors that are significantly affected by diet: oxygen and fuel. Obviously on a ketogenic diet, the primary fuel will be fat rather than glucose (or glycogen, but I'll discuss that in a moment). In this context, the primary theoretical benefit of fat over glucose is its energy density - it yields more ATP and therefore energy per unit than glucose. This fact also affects the oxygen part of the equation. Since it's more energy-efficient, it's more oxygen-efficient, because more energy is produced per unit of oxygen.
That all seems pretty straightforward - fat wins! But it's complicated by the issue of availability. On a high-carbohydrate diet, glycogen is stored in the muscles, and it provides a very readily available fuel, delivered very quickly during high-intensity exercise. A ketogenic diet quickly depletes glycogen stores, so once the athlete has "adapted" to the diet, he likely has little to no intramuscular glycogen available for use as fuel. He does, however, have considerably more intramuscular triglycerides to replace it, and these are readily available as fuel. Trained athletes are able to metabolize (oxidize) these fat stores more effectively than non-athletes, so we would expect different results before and after the diet change, depending on the runner's level of fitness.
Another critical question is what happens when oxygen can no longer be delivered to the muscle cells quickly enough to keep up with demand. At the beginning of a sprint, this is no problem, but at some point that differs from runner to runner, an anaerobic system takes over which burns fuel without oxygen. If the fuel is glucose, cells use lactic acid fermentation to extract energy from the fuel without the benefit of oxygen. This results in a buildup of lactic acid in the muscles, which contributes to fatigue and the familiar burning and soreness associated with high-intensity exercise. So what happens when you're burning fat instead of glucose, and you run out of oxygen? If anyone knows the answer to that, I haven't found it. There's some reason to believe creatine is involved, but at this point it's not clear, at least to me. That dieters don't fall over from muscle failure is evidence that there's an anaerobic alternative to fatty acid metabolism, but I haven't been able to identify it.
So on the question of high-intensity exercise, the scholarly research is scarce and flawed, and the theory suffers from the unknown factor of anaerobic lipid metabolism. My prediction, then, is based largely on an anthropological idea:
As I've discussed more than once, carbohydrates as a substantial portion of the human diet are relatively new. Prior to the invention of agriculture, it seems that most humans spent most of their lives in ketosis, fed by meals whose acquisition likely involved a lot of sprinting. It stands to reason, then, that humans are well-suited to sprinting while in ketosis, and I'm prepared to allow this conclusion to fill in some of the blanks and predict that after a period of adaptation, a sprinter should outperform his non-ketogenic self.
Extreme Endurance ExerciseHere there seems to be no scholarly research at all. The closest thing I've found is an account of an arctic search and rescue expedition involving involuntary ketosis. Food supplies had been depleted, and food was only available from hunting and fishing. After several days of despair attributed to the adaptation period, things of course turned around for the rescue party. The leader of this expedition, Lt. Frederick Schwatka, describes it like so:
When first thrown wholly upon a diet of reindeer meat, it seems inadequate to properly nourish the system, and there is an apparent weakness and inability to perform severe exertive fatiguing journeys. But this soon passes away in the course of two or three weeks.
And the author of the paper shares his insight into the effect of this diet on men carrying a lot of heavy gear and encumbered by arctic-weather clothing:
That Schwatka was not impaired by his prolonged experience eating meat and fat is evidenced by his diary entry for the period 12–14 March 1880, during which he and an Inuit companion walked the last 65 miles in less than 48 hours to make a scheduled rendezvous with a whaling ship and complete his journey home.
Non-athletes, working under some of the least favorable conditions imaginable, presumably carrying a heavy load and wearing heavy, cumbersome clothing, covering 65 miles in less than two days is impressive.
However, I have my doubts that this translates very well into dietary recommendations for marathoners. Marathoners of course tend to be very lean, and eating a ketogenic diet won't change that. The amount of fat available for use as fuel is likely to be insufficient, though there's not a lot of evidence either way on that. The energy efficiency of fat metabolism may, thermodynamically speaking, allow for a large, fatty meal to see an athlete successfully through a marathon, but I'm not prepared to recommend that anyone try it.