Ketone drinks: performance, recovery and cognition
In this investigation
- What would make them useful here?
- The compound and feeding context matter
- An immediate performance boost is unreliable
- The promising training trial deserves attention
- Recovery results do not support a general shortcut
- Cognition and sleep are narrower than the headlines
- What I would put ahead of a ketone drink
What would make them useful here?
For my current food and training routine, I see no clear reason to add ketone drinks. The interesting question is whether they improve a result I care about while I am adequately fed: a better workout, better recovery or clearer thinking. Raised blood ketones do not establish a training benefit; trial results differ. This is a research topic to watch rather than another missing nutrient to buy.
The compound and feeding context matter
Ketone drinks can raise blood β-hydroxybutyrate (BHB) despite carbohydrate feeding. In resting volunteers, a meal lowered the monoester’s peak BHB; salt and ester differed metabolically. Human metabolism experiments
| Form | What remains distinct |
|---|---|
| BHB monoester | (R)-3-hydroxybutyl (R)-3-hydroxybutyrate. |
| BHB salts | The tested sodium/potassium mixture contained D- and L-BHB, with different blood persistence. |
| Acetoacetate diester | A different compound, tested separately below. |
The metabolic comparison identifies salt/ester exposures; the cycling experiment identifies the diester.
An immediate performance boost is unreliable
The following trials ask different questions; their percentages should not be averaged casually.
| Trial and context | What changed | Practical limit |
|---|---|---|
| 2023: 23 trained cyclists, monoester before a 20-minute time trial | Mean power was 2.4% lower, with a 95% confidence interval of 0.6–4.1% lower, despite raised BHB. Original trial | An identical monoester can impair performance in a different context. |
| 2026: eight trained men, carbohydrate-loaded, then three hours cycling with 120 g carbohydrate/hour | The monoester lowered glucose and oxidation of ingested carbohydrate. Subsequent exercise capacity was not better than carbohydrate alone. Original trial | Fuel changes are not necessarily advantageous; this high feeding rate is not my short-workout requirement. |
| 2023: 28 WorldTour cyclists, monoester added against a background of bicarbonate in both conditions | The initial eight-minute test improved about 3.1%; the abstract’s benefit concerns the initial phase, before 4.5 hours riding. Original trial | Specific combined context and initial test; lasting stage benefit unestablished. |
| 2017: ten healthy men, ketone salts after an overnight fast | High-intensity time-trial power was about 7% lower while estimated fat oxidation rose. Original salt trial | Salt and fasting differ from the fed monoester question. |
A separate acetoacetate-diester experiment in professional cyclists reported slower performance and gastrointestinal discomfort in all ten finishers; one additional rider could not complete the test after severe symptoms. That matters for the tested diester, and should not become a claim that every ketone formulation has those effects. Original diester trial
The promising training trial deserves attention
In a 2026 eight-week trial, 28 of 30 trained men completed supervised cycling with postexercise/presleep monoester or active, calorie-matched medium-chain triglyceride (MCT) control. Both groups received carbohydrate during longer sessions and carbohydrate/protein recovery feeding. Week-seven 30-minute power was 302 versus 291 W: 11 W higher (95% CI 4–18); a smaller difference persisted after taper. Citrate synthase activity favored ketones; cardiac and subjective recovery outcomes did not. Original study
Each 25 g monoester dose added 489 kJ (~117 kcal). Performance followups used Šidák correction; metabolomics used false-discovery-rate control, without one correction across all outcomes. Markers do not establish mechanism or tendon repair. Methods. First registry submission followed the listed primary completion, limiting advance-selection claims. Registry.
Recovery results do not support a general shortcut
In 2026, 18 recreational runners used a high daily monoester exposure or a calorie-matched placebo for 31 days. Their 5 km performance did not improve relative to placebo. Selected cognitive results differed, but mood and quality of life did not. The running test followed recent feeding and assigned supplement ingestion; separate metabolic testing used overnight fasting without an acute dose. These assessments should not be treated as the same exposure. This single-blind study received college and KetoneAid support, with ketone-related commercial interests disclosed by several authors. Original runner trial
Another 2026 trial, published August 27, tested 18 well-trained cyclists during an eight-day simulated stage race. Postexercise and presleep monoester did not clearly improve performance, sleep, cognition or recovery. Time-trial performance deteriorated across the race, and the supplement did not prevent that decline. The abstract also reports worse mechanical efficiency and increased fat oxidation in the ketone group. Its small size does not settle every possible context. Original stage-race trial
A 2024 crossover study in nine men tested ketones plus oral carbohydrate during four hours of recovery after glycogen-lowering exercise. BHB and insulin changed, but subsequent 20 km and 5 km cycling performance did not significantly improve. Muscle glycogen was not measured, so the study cannot establish exactly why. Original recovery trial
Earlier 2019 overload research in 18 men reported better tolerable training load and a late-session endurance test, but the groups also diverged in spontaneous food intake and energy balance. That prevents a clean conclusion that ketone signaling alone caused the result. Original overload study A contemporaneous methodological critique also disputed the claim that this established prevention of nonfunctional overreaching, which requires an appropriate individual performance and recovery trajectory. Original critique
Cognition and sleep are narrower than the headlines
A July 2026 crossover trial in 25 adults found that monoester helped preserve selected reaction-time tasks after experimentally induced mental fatigue. The largest reported improvements involved ketones with caffeine. There was no caffeine-only arm in the reported three-condition design, so it cannot establish that adding ketones beats caffeine alone. Subjective fatigue and workload did not differ by supplement. These laboratory tasks do not measure my work productivity or durable cognitive health. Original mental-fatigue trial
A separate experiment, published online in December 2025, studied nine young active men after carbohydrate feeding and moderate cycling. A broad cognitive battery showed mostly null responses. There was an interaction suggesting preserved global motion perception, but no corresponding general improvement in memory or sustained attention. With many tested outcomes and a small sample, that selected result needs replication. The article’s companion physiology paper concerns the same participants, not an additional independent cognitive trial. Original cognitive-battery study
In 12 trained women, a monoester/carbohydrate condition improved selected postexercise cognitive measures while leaving cycling time essentially unchanged. This adds a different population and task, rather than establishing the same effect in me. Original female-athlete trial
Sleep also has a positive but narrow signal: a 2023 crossover study in ten cyclists found better measured sleep efficiency after a demanding day with exercise ending one hour before bed. That does not establish ketones as a general sleep aid, particularly beside the null sleep result during the 2026 stage simulation. Original sleep experiment
What I would put ahead of a ketone drink
My reported routine includes short running sessions and lifting, not the supervised endurance workload in the eight-week trial. The present priority is a food routine I enjoy, adequate ordinary fueling and recoverable training. The energy and recovery investigation explains why a supplement biomarker cannot replace that foundation.
If relevant later, I would name the compound and desired outcome. Calories and stomach tolerance matter. The tested salt’s sodium/potassium exposure cannot specify every product’s mineral content. Salt identity.
No personal ketone use is reported in my log. These trials do not establish long-term benefit, safety or shoulder treatment. Caffeine has its own dose and sleep context, and persistent pain belongs in the recovery discussion.