Abstract-Population densities of six dominant macroinvertebrate taxa (mayfly nymphs Jappa kutera, Atalophlebia sp., Tasmanocoenis sp., Baetis sp. and the caddisfly larvae Cheumatopsyche sp. and Ecnomus sp.) were negatively correlated to total endosulfan concentrations in the Namoi River (New South Wales, Australia) in the 1995/96 and 1997/98 cotton growing seasons (NovemberMarch). Total endosulfan concentrations measured in solvent-filled polyethylene bags (passive samplers) at the exposed sites correlated with rainfall, suggesting that endosulfan entered the riverine environment in runoff from land. At the start of both surveys, in November 1995 and November 1997, there was no significant difference (p Ͻ 0.05) between the reference and exposed sites for both total endosulfan concentrations and mean population densities of the combined study taxa. This indicates distance downstream, which was negatively correlated with exposure, was not an important variable in explaining the significantly higher population densities in other months at the reference sites compared with those at sites subsequently exposed to increased (10-25-fold) total endosulfan concentrations. The multivariate analysis of the 1997/98 data with the principal response curves (PRC) method indicated that endosulfan explained a significant proportion (25%) of the variation in the total macroinvertebrate community. Principal components analysis (PCA) indicated other covariables were involved, including river discharge. River discharge was positively correlated to increased densities of the mayfly taxa at the reference sites in the 1995/96 survey, but these correlations were reduced to near zero, except for Baetis sp., in the 1997/98 survey.