Water quality in the Pleasant Grove Spring karst groundwater basin, Logan County, Kentucky, was monitored to determine the effectiveness of best management practices (BMPs) in protecting karst aquifers. Ninety-two percent of the 4,069-ha (10,054-acre) watershed is used for agriculture. Water-quality monitoring began in October 1992 and ended in November 1998. By the fall of 1995 approximately 72% of the watershed was enrolled in BMPs sponsored by the US Department of Agriculture Water Quality Incentive Program (WQIP). Pre-BMP nitrate-nitrogen concentration averaged 4.65 mg/1. The median total suspended solids concentration was 127 mg/1. The median triazine concentration measured by immunosorbent assay was 1.44 ??tg/l. Median bacteria counts were 418 colonies per 100 ml (col/100 ml) for fecal coliform and 540 col/100 ml for fecal streptococci. Post-BMP, the average nitrate-nitrogen concentration was 4.74 mg/1. The median total suspended solids concentration was 47.8 mg/1. The median triazine concentration for the post-BMP period was 1.48 ??g/1. The median fecal coliform count increased to 432 col/100 ml after BMP implementation, but the median fecal streptococci count decreased to 441 col/100 ml. The pre- and post-BMP water quality was statistically evaluated by comparing the annual mass flux, annual descriptive statistics, and population of analyses for the two periods. Nitrate-nitrogen concentration was unchanged. Increases in atrazine-equivalent flux and triazine geometric averages were not statistically significant. Total suspended solids concentration decreased slightly, whereas orthophosphate concentration increased slightly. Fecal streptococci counts were reduced. The BMPs were only partially successful because the types available and the rules for participation resulted in less effective BMPs being chosen. Future BMP programs in karst areas should emphasize buffer strips around sinkholes, excluding livestock from streams and karst windows, and withdrawing land from production.