Atrazine and two of its metabolites, deethylatrazine (DEA) and deisopropylatrazine (DIA), are frequently detected in surface runoff. Although their health and environmental effects may be similar to that of atrazine and ratios of their concentrations are useful in delineating sources of contamination, there have been few long-term studies of the factors affecting their losses in runoff. Therefore, losses of atrazine, DEA, and DIA were monitored for six years in runoff from seven small (0.45-0.79 ha) watersheds under three tillage practices. Weather year and the timing of runoff-producing rainfall had a greater effect on atrazine, DEA, and DIA concentrations and losses than did tillage practice. DEA was the most frequently detected metabolite with an average concentration in the year of atrazine application, of 2.5 microg L(-1) compared to 0.7 microg L(-1) for DIA. Atrazine exceeded its 3 /g L(-1) maximum contaminant level (MCL) up to 100 days after application. DEA and DIA exceeded the atrazine MCL up to 50 days after atrazine application; thus, failure to monitor their presence may result in underestimation of the environmental impact of atrazine usage. The molar concentration ratio of DEA to atrazine (DAR) was affected by tillage treatment, weather year, and possibly soil type. These factors may need to be taken into account when DAR is used as an index of atrazine movement. The ratio of DIA to DEA (D2R) was fairly constant and should be useful in determining the parent compounds contributing DIA to surface waters.