We assessed the effect of sampling schedule on estimates of diet composition of an age-II+ cohort of largemouth bassMicropterus salmoidesin a small lake in Michigan's upper peninsula. Our objectives were to determine the effects of sampling frequency on diet variance and on estimated prey consumption by the fish population. Sampling frequency ranged from 2 to 30 samples per 45-d sampling period in summer 1987. We determined the mean variance of an index of relative importance for each of six prey types as a function of sampling frequency. Estimates of mean rates of total population food consumption (based on percent wet weights of each prey type) were calculated with a generalized bioenergetics model. Feeding on each of three prey items (i.e., zooplankton, chironomids, and trichopterans) fluctuated irregularly over time, and intervals of 3–18 d between feeding bouts were determined from the data. Sampling frequency influenced the accuracy of consumption rate estimates. A minimum sampling frequency of three times per 45 d was necessary for accurate estimates of consumption rates forChaoborusspp., for odonate naiads, and possibly for ephemeropterans. However, accurate estimates of consumption rates for zooplankton (Daphniaspp. andHolopedium gibberum), trichopterans, and chironomid pupae required more frequent sampling. Increased sampling did not decrease the intersample variance, but the error in estimates of consumption rates decreased with increased sampling frequency.