In this study, we investigated use of microcosms to supplement field studies for establishing the size of wetlands required to mitigate nitrate pollution in agricultural watersheds. Wetlands investigated in this study were located in San Joaquin Valley (California, USA) and demonstrated mean nitrate-nitrogen mass removal efficiencies ranging between 10 and 34%. Mean areal nitrate removal rates (J) ranged from 142 to 380 mg-N m(-2)d(-1). First-order rate constants determined from field data had a high variance, with confidence intervals greater than 57% of mean values. Sediments and rooted plants from one site were placed in a flow-through microcosm and measurements of nitrate removal kinetics were made and compared with field results. The apparent half-saturation constant (K-m) and maximum removal rate (J(max)) for nitrate-nitrogen were 43.8 mg/L and 4.11 g m(-2)d(-1) in the microcosm. The first-order rate constant from the microcosm (10.4 cm d(-1)) was in close agreement with the value for the field site (11.9 cm d(-1)) and had a confidence interval of less than 16%. Using this improved first-order rate constant, it was determined that between 1.3 and 3.6% of the land in the watershed should be managed as mitigation wetland, with the area required dependent on the level of nitrate reduction desired and how closely the wetland design approximates plug-flow.