The effects of 10 sampling strategies on estimates of annual mean concentrations of the herbicides atrazine, alachlor, and cyanazine in selected midwestern rivers were tested. The accuracy of the strategies was computed by comparing time-weighted annual mean herbicide concentrations calculated from water samples collected from 17 locations on midwestern rivers, with simulated annual mean concentrations calculated for each sampling strategy, using Monte Carlo simulations. Monthly sampling was the most accurate strategy tested. The U.S. Environmental Protection Agency requires quarterly sampling for municipalities using surface water as a source of drinking water. Due to the seasonality of herbicide occurrence and transport, quarterly sampling underestimates annual mean herbicide concentrations in over 40% of the simulations. Three of the strategies tested showed that, relative to quarterly sampling, a more accurate representation of annual mean concentrations could be obtained by sampling more frequently during spring and early summer runoff and assuming zero herbicide concentration during late summer and winter months.