Microbial contamination is a major concern for drinking water worldwide. Many monitoring protocols that use one or very few samples are inadequate and introduce a very large margin of error. An intensive sampling program needs to be conducted to characterize the Escherichia coli concentrations of a source water stream prior to establishing a monitoring program so that the sample frequency can be determined statistically, based on an acceptable margin of error. Developing meaningful monitoring programs for managing bacterial water quality is dependant on scientific data that determine the bacterial sources. In this study, three streams from drinking water watersheds were sampled every 15 min over a 24 h period on three different days to determine the concentrations of E. coli and to identify their sources, using ribosomal RNA finger printing (ribotyping). The concentrations of E. coli varied throughout the day in each of the three streams. Ribotyping identified many different animal sources of E. coli in the samples. The sources of E. coli varied significantly with stream (P < 0.001, df = 16). The development of monitoring programs for watersheds needs to consider the watershed, and care needs to be taken in selecting appropriate sample sites, sampling regime, and number of samples taken during each sampling period. This note provides a prescription for the development of monitoring programs for watersheds.