The influence of specific training on benefits from caffeine (Caf) ingestion was examined during a sprint test in a group of highly trained swimmers (T) and compared with the response of a group of untrained occasional swimmers (UT). Seven T and seven UT subjects swam freestyle two randomly assigned 2 x 100 m distances, at maximal speed and separated by 20 min of passive recovery, once after Caf (250 mg) and once after placebo (Pla) ingestion. Anaerobic capacity was assessed by the mean velocity (meters per second) during each 100 m and blood was sampled from the fingertip just before and 1, 3, 5, 7, and 9 min after each 100 m for resting and maximal blood lactate concentration ([la-]b,max) determination. The [la-]bmax was significantly enhanced by Caf in both T and UT subjects (P less than 0.01). However, only T subjects exhibited significant improvement in their swimming velocity (P less than 0.01) after Caf or any significant impairment during the second 100 m. In light of these results, it appears that specific training is necessary to benefit from the metabolic adaptations induced by Caf during supramaximal exercise requiring a high anaerobic capacity.