Field trials were established in 2005 and continued in 2006 to evaluate a conventional broadcast herbicide sprayer compared to a variable spray (sensor-activated) weed-sensing sprayer (WSS). The computer-based Herbicide Application Decision Support System (WebHADSS) was used to determine a portion of the herbicides applied (based on herbicide efficacy and economics). Weed control, herbicide usage, crop yield, and net returns were compared across treatments. The broadcast applications were usually the most effective at controlling weeds. A PPI herbicide did not always improve weed control compared to treatments in which no PPI herbicide was applied. Variable treatments used less herbicide than the broadcast system in both years. Cotton lint yields in broadcast applications were similar to the weed-free check in both years of the study. Variable treatments often provided equivalent net returns (gross yield revenue less weed control cost) to the broadcast treatments. Although herbicide savings were observed in the variable treatments when compared to a broadcast system, a reduction in weed control was observed, indicating the need for future improvements of this system. A site-specific weed management program used in conjunction with WebHADSS may have potential in cotton production systems in the Texas Southern High Plains where weed densities are low.