Samples from two streams (Kisco River and the Middle Branch of the Croton River) in the Croton Reservoir system in south-eastern New York State, USA were sampled from May 2000 through to February 2001 in order to document the effect of land use, streamflow and seasonal patterns of application on pesticide concentrations in runoff from developed watersheds. Many of the pesticides detected most commonly in this study are generally used in developed areas, and particularly on turfgrass. Pesticide concentrations were generally higher, and the numbers of compounds were generally larger, in samples from the Kisco River than in samples from the Middle Branch, probably because the Kisco River drainage has a greater population density and is more extensively developed. Four pesticides (2,4-D, 2,4-D-methyl, dicamba and metalaxyl) were detected in at least one sample from the Kisco River at a concentration > 1 ??g litre-1, and no pesticides were detected at concentrations >0.4 ??g litre-1 in Middle Branch samples. No human-health-based water-quality standards were exceeded by samples from either site in this study, but samples from the Kisco River contained four insecticides (carbaryl, chlorpyrifos, diazinon and malathion) and one herbicide (2,4-D) in concentrations that exceeded water quality criteria for the protection of aquatic life. The highest concentrations of most compounds occurred during stormflows in both streams in June, September and December, 2000. The lowest concentrations of most compounds at both sites occurred during baseflows from October 2000 through February 2001, even though the concentrations of many compounds increased substantially at the Kisco River site during stormflows in November and December. Detailed data on the variability of pesticide concentrations during stormflows indicate that there may be two sources of pesticides in the Kisco River watershed: (1) elevated concentrations of pesticides during peak flows that occur early in stormflows likely reflect runoff from paved areas, and (2) elevated concentrations during peak flows that occur later in stormflows from areas with lesser amounts of pavement. Data from the Kisco River indicate that the relation between storm discharge and pesticide concentrations varies among compounds, in part because of variation in seasonal application patterns. These variations in the timing of application result in not all stormflows producing increased concentrations of pesticides. Overall, these results indicate the importance of stormflow sampling throughout the year in assessing pesticide late and transport in urbanized, developed areas.