Concentrations of mecoprop, dichlorprop, and metolachlor were investigated in 393 samples from Ontario streams in 2003-2004. Maximum concentrations of mecoprop, dichlorprop, and metolachlor were 1900, 6.62, and 1560 ng L(-1), respectively in 2003, and 103,000, 110, and 5290 ng L(-1), respectively, in 2004. Concentrations of metolachlor in agricultural watersheds were significantly (p < 0.0001) higher than those in urban watersheds, whereas concentrations of mecoprop did not differ (p > 0.1) between the two watershed types. Enantiomer fractions (EFs) of the acid herbicide mecoprop ranged from 0.236 to 0.928 and for dichlorprop EFs ranged from 0.152 to 0.549. EFs of mecoprop did not differ significantly (p > 0.1) between agricultural and urban land usage. Mecoprop EFs > or = 0.5 were found in 54% of samples, a situation expected during the transition years when racemic mecoprop was replaced by mecoprop-P, consisting of only the R(+) enantiomer. However, EFs < 0.5 were found in 46% of samples, which suggests preferential degradation of R(+) or conversion of R(+) to S(-). The four stereopluisomers of metolachlor were separated by HPLC and are reported as the ratio of the two herbicidally active stereoisomers (aS,1'S and aR,1'S) to the two inactive stereoisomers (aS,1'R and aR,1'R). The average S/R of metolachlor in streams was 6.73 +/- 2.28 and ranged from 1.08 to 12.7. Samples with higher concentrations of metolachlor had S/R ratios similar to the S/R of S-metolachlor (enriched in aS,1'S and aR,1'S stereoisomers) whereas samples with lower concentrations display lower S/R ratios with greater variability, suggesting mixed input of racemic and S-metolachlor or stereoselective degradation. S/R values were significantly higher (p < 0.001) in agricultural than urban watersheds. Streamwater from row crop areas showed higher S/R ratios than streamwater from fruit growing areas (p = 0.036).