An algorithm was developed for use in a geographic information system (GIS) to model the surface movement of herbicide in response to a rainfall event as modulated by slope, soil, management practices, and time of herbicide application. This algorithm was implemented in the GIS software Geographic Resource Analysis Support System (GRASS) and uses as submodels the Natural Resources Conservation Service (NRCS) curve number procedure, the Universal Soil Loss Equation (USLE), and the pesticide submodel from the model Chemicals, Runoff, and Erosion from Agricultural Management Systems (CREAMS). The algorithm estimates the loss of pesticide from field areas, runoff flow patterns, and the accumulation of pesticide downslope in response to a rainfall event. The simulated movement of atrazine, cyanazine, and alachlor was studied under hypothetical management scenarios in the Lake Pittsfield watershed in Pike Co., IL. Tillage for the simulation was by moldboard plow. An alternate no-till scenario was simulated to test tillage effect on atrazine movement. Herbicides were applied either PPI, PRE, POST, or early preplant for no-till (treated as same application time as PPI but without incorporation). The experiment was designed to incorporate timing of application as a management factor from the standpoint of a single rain event on May 16. The results used for comparison were data from 1 d after POST application, 15 d after PRE application and 30 d after PPI application. The algorithm showed that areas of greater herbicide risk can be located within a watershed and that the effect of alternative management practices can be evaluated using a GIS.