Juvenile rainbow trout (approximately 6 g) were exercised to exhaustion in two 5 min bouts given 6 h apart. Resting levels of whole-body lactate and glycogen were restored prior to the second bout. The rate of O2 consumption increased about threefold 5 min after each bout of exercise, while recovery time decreased from 4 h after the first bout to 2–3 h after the second. The excess post-exercise oxygen consumption, i.e. ‘oxygen debt’, was significantly reduced by 40% after the second exercise bout, despite almost identical rates of lactate clearance and glycogen resynthesis. The rates of CO2 and ammonia excretion increased sixfold and threefold, and recovery times decreased from 4–6 h to 3 h and from 3 h to 1.5 h, respectively. After the first bout, whole-body lactate levels peaked at 5 min post-exercise at about 8.5 times pre-exercise levels. After the second bout, lactate levels peaked at 0 min post-exercise and fell more rapidly during recovery. Whole-body glycogen levels decreased by 70% and 80% and ATP levels decreased by 75% and 65% after the first and second bouts, respectively, while glucose levels increased about 1.5-fold immediately after both bouts. Creatine phosphate levels decreased by 70% and 80% after the first and second bouts, respectively. After 6 h of recovery, creatine phosphate levels were higher after the second bout than after the first. These findings suggest that exhaustive exercise may cause a ‘non-specific’ increase in metabolic rate not directly related to the processing of metabolites, which is reduced upon a subsequent exercise bout. This is in contrast with the classical ‘oxygen debt hypothesis’, which states that the oxygen debt and lactate clearance are linked. Furthermore, it appears that two sequential exercise bouts are sufficient to induce a ‘training effect’, i.e. improved rates of metabolic recovery.