The belief that high-carbohydrate diets enhance training capacity (mean power output) has been extrapolated from studies that have varied dietary carbohydrate over a few days and measured muscle glycogen but did not assess power output during training. We hypothesized that a high-carbohydrate (HI) diet (10 g.kg body mass-1.day-1) would promote greater muscle glycogen content and greater mean power output during training than a moderate-carbohydrate (MOD) diet (5 g.kg body mass-1.day-1) over 4 wk of intense twice-daily rowing training. Dietary protein intake was 2 g.kg body mass-1.day-1, and fat intake was adjusted to maintain body mass. Twelve male and 10 female collegiate rowers were randomly assigned to the treatment groups. Training was 40 min at 70% peak O2 consumption (VO2) (A.M.) and either three 2,500-m time trials to assess power output or interval training at 70-90% peak VO2 (P.M.). Mean daily training was 65 min at 70% peak VO2 and 38 min at greater than or equal to 90% peak VO2. Mean muscle glycogen content increased 65% in the HI group (P less than 0.05) but remained constant at 119 mmol/kg in the MOD group over the 4 wk. Mean power output in time trials increased 10.7 and 1.6% after 4 wk in the HI and MOD groups, respectively (P less than 0.05). We conclude that a diet with 10 g carbohydrate.kg body mass-1.day-1 promotes greater muscle glycogen content and greater power output during training than a diet containing 5 g carbohydrate.kg body mass-1.day-1 over 4 wk of intense twice-daily rowing training.(ABSTRACT TRUNCATED AT 250 WORDS)