The present investigation examined the influence of dietary carbohydrate (CHO) on the performance of supramaximal intermittent exercise. A group of 14 moderately trained male students [mean age 21.0 (SD 1.6) years] completed a maximal oxygen consumption test and two identical 'maximal interval tests' (MIT1 and MIT2) over a 10-day experimental period. Each MIT involved five 60-s all-out periods of cycling (against 0.736 N.kg-1 body mass), with each period separated by 5 min of passive recovery. All subjects consumed a moderate CHO diet for 3 days preceding MIT1 (55.3% of energy intake as CHO) and were then randomly assigned to either a high CHO (83%), moderate CHO (58%) or low CHO (12%) diet for the 3 days separating MIT1 and MIT2. All food and drink consumed during the experimental period was weighed and recorded for later dietary analysis. Measurements of work done, exercise oxygen consumption (VO2), venous blood pH, plasma lactate and plasma glucose concentrations were compared between interval tests. Independent Student's t-tests revealed that the 5.6% increase in total work done recorded by the high CHO group and the 2.3% increase by the moderate CHO group between MIT1 and MIT2, were significantly different to the 5.4% decrease in performance recorded for the low CHO group (P < 0.001 and P < 0.01, respectively). No significant differences in performance changes was found between the high and moderate CHO groups; 10 subjects in each of these groups would have been necessary to attain significance at the 0.05 level using the same procedures.(ABSTRACT TRUNCATED AT 250 WORDS)