Selective sprayer units are designed to spray the plants between crop rows (weeds) and not spray over bare ground. These systems have proven effective at reducing herbicide usage with minimal impact on weed control. Limited research has been reported on the accuracy of these systems. This paper will address a technique for evaluating overall system accuracy. The University of Tennessee developed an automated mapping system that allows weed maps to be generated in conjunction with selective spray operations. This system records an integrated nozzle activity (% on time) with an associated GPS coordinate every second. Maps created with this system can then be used to account for site-specific placement of herbicides, track temporal changes in weeds, and provide a target layer for precision pre-emerge herbicide applications. Accuracy of these weed maps are strongly influenced by the sensing unit's ability to differentiate between weeds and soil, response time of the fluid delivery system, and GPS accuracy. There is no automated technique for field evaluation of these units. A video documentation system has been developed and tested that provides feedback on the performance of these mapping systems. A high resolution video camera records in real-time the area that is being sensed by the weed sensor. Custom analysis software was developed to analyze the video for weeds sprayed, weeds missed, and spray events not triggered by weeds. A measure of weed clustering is also reported. Various weed functional types, growth stages, and spatial distributions under field conditions were used in the evaluation. The overall evaluation process and results for one selective spray unit (WeedSeeker) will be reported.