Water samples were taken four times per year and bed-sediment samples two times per year during 1975-80 at 160 to 180 stations on major rivers of the United States. Samples were analyzed for 18 insecticides and 4 herbicides, which together accounted for about one-third of the total amount of all pesticides applied to major crops during 1975-80. Fewer than 10 percent of almost 3,000 water samples and fewer than 20 percent of almost 1,000 bed-sediment samples contained reportable concentrations of any of the compounds. The patterns of detection result from a combination of widely variable detection capabilities, chemical properties, and use. Most detections in water samples were of relatively persistent yet soluble compounds: atrazine (4.8 percent of samples), diazinon (1.2), and lindane (1.1). Most detections in bed-sediment samples were of the hydrophobic and persistent insecticides: DDE (17 percent of samples), DDD (12), dieldrin (12), chlordane (9.9), and DDT (8.5). Only for atrazine in water, and for DDE, DDD, DDT, and chlordane in bed sediments, were geographic patterns of detection correlated (pH<0.10) with use on farms. Detections of organochlorine insecticides in both water and bed sediments appear to have erratically but gradually decreased during 1975-80. For the 1975-79 period, more stations had downtrends than had uptrends in bed-sediment levels of organochlorines. No clear trends were evident in concentrations of organophosphate insecticides or herbicides in either water or bed sediments. Findings suggest that future pesticide monitoring efforts must be responsive to changes in pesticides used and to geographic patterns of use. Different types of monitoring approaches are necesssary for chemicals having different chemical and physical properties. Before an effective dynamic monitoring effort can be designed, however, selected case studies are needed to characterize and refine sampling and analytical capabilities for different types of chemicals, river environments, and sample types.