Creatine: exposure, timing and what actually improved

Updated 11 Oct 2026

The useful question for my training

My reported training includes short sprints and lifting with warm-up sets followed by one hard set per exercise. Creatine remains a plausible optional experiment for repeated high-intensity work. No use is recorded here. The practical question is whether it improves a repeatable training outcome enough to justify taking it consistently—not whether a new paper has discovered an exact minute to swallow it.

The existing performance-supplements investigation covers the broader case for creatine monohydrate. This page examines exposure and measurement more closely. Study doses below describe experiments; they do not establish a personal regimen.

A dose in the bloodstream and a muscle store are different things

In a classic study of 31 men, 20 g/day for six days increased total muscle creatine by about 20%. A smaller 3 g/day protocol reached a similar increase more gradually, over 28 days. After loading stopped without maintenance, stores declined gradually and were no different from baseline after approximately 30 days. These were muscle measurements, not proof that every performance outcome improves equally under either protocol. Hultman et al., 1996.

This makes two distinctions useful. Consistent exposure over time is a different question from a single pre-workout dose. And a crossover study must consider what remains from the previous condition: clearing a transient blood concentration does not automatically mean muscle stores have returned to baseline.

A recent timing trial found no winning window

A June 2026 trial randomized 37 adults and analyzed 27 completers after 16 weeks of resistance training. Creatine was 5 g on training days, immediately before or after training, compared with placebo. It was not 5 g every day. Neither timing produced a significant additional improvement in body composition, ultrasound muscle thickness or performance. Several performance measures improved across the groups with training. Mills et al., 2026.

The result deserves a bounded interpretation. The analyzed groups were small, adherence varied, muscle creatine was not measured, and training used multiple sets across eight exercises. It cannot show that all consistent creatine protocols fail, nor demonstrate that increasing the dose would fix this particular null result. It does weaken the case for treating before-versus-after timing as a priority for my current routine. Trial methods and limitations.

A separate 11-man acute crossover pilot tested 0.1 g/kg around exercise. The pre-exercise condition exceeded during-exercise for squat load. For bench press, both pre- and during-exercise creatine—and the no-supplement control—exceeded placebo. Placebo matched only the two-hour pre-exercise timing. The prescribed fixed-load, fixed-repetition session leaves unclear how variable total lifted load was obtained. The authors called the findings exploratory; this is insufficient to establish a timing rule. Ben Maaoui et al., 2026.

Better total work does not mean every fatigue measure improves

In a December 2025 crossover study of 14 active men, seven days of 20 g/day increased distance across six 30-second shuttles with 35-second rests: medians 762.5 versus 662.5 m. Best-bout distance improved; fatigue index, percentage decline and perceived exertion did not significantly differ. These bouts differ from my 10-second sprints. Ben Maaoui et al., 2025.

The 14-day washout lacked muscle measurements, leaving carryover uncertain. Both papers share a registry ID; unresolved participant overlap prevents counting them as independent samples. Loading study, acute pilot.

Early lean mass needs an independent performance check

An early scale or DXA change needs interpretation alongside performance. The wash-in trial account separates supplement-only and training phases; it does not identify how much change was water or new contractile muscle.

Sleep quality and cognition require their own evidence

The loading study improved subjective sleep quality; actigraphy duration, efficiency and latency did not significantly differ. Soreness was higher at 24 hours (P = .016), without other significant post-exercise soreness or perceived-recovery differences. Pre-testing perceived recovery was lower. This does not establish improved recovery or insomnia treatment. Sleep and recovery results.

A separate April 2026 trial of 29 adults tested a single 0.2 g/kg dose during 21 hours of wakefulness. The pooled logic result met the stated Bonferroni threshold (P = .005; threshold .0063). Numerical, language-processing and vigilance-dispersion findings were only nominally significant and did not meet that threshold. Not all tests improved. The study was retrospectively registered and reported no physical adverse effects in its small sample. It did not test ordinary rested work or chronic sleep restriction. Gordji-Nejad et al., 2026.

The rested-performance evidence is less decisive. A larger preregistered crossover trial used 5 g/day for six weeks per condition: neither working memory nor reasoning reached conventional statistical significance, and eight exploratory tests showed no clear improvement. Bayesian analyses left room for a small benefit. Vegetarians did not benefit more than omnivores in that trial. Sandkühler et al., 2023.

These findings justify further research without making strength evidence prove cognitive enhancement. Preserving sleep remains a separate decision, especially when caffeine and late training are involved.

Tolerance belongs beside performance

In 59 male soccer players, taking 10 g at once for 28 days produced more reported diarrhea than the same daily amount split into two 5 g doses: 55.6% versus 28.6%; placebo was 35%. The percentages come from one small trial and are not universal expected rates. They show why the size of a single serving matters alongside the daily total. Ostojic and Ahmetovic, 2008.

The larger cognitive trial also recorded more reported side effects with creatine than placebo (17% versus 4%), mostly mild reports; no participant stopped because of an adverse event. This is more informative than assuming tolerability from a study that did not systematically report it. Sandkühler trial, adverse-event table.

What would make an experiment useful here?

My current priority is a stable, tolerable routine and an endpoint I can repeat. If creatine is later adopted, the log should record actual exposure, missed days, gastrointestinal symptoms, body-weight trend and a consistent performance measure. For track work, distinguish fastest-bout performance from total work and the drop across bouts. For lifting, record load, repetitions and rest under the same exercise setup.

Compare enough sessions to understand ordinary variability, while recognizing that ongoing training improvement and expectation still confound a personal experiment. A good week after starting is a useful observation, not a randomized trial. Shoulder symptoms remain their own issue; an extra repetition or less perceived effort cannot establish that loading the shoulder is appropriate.

Back to investigations