The influence of different sampling strategies on characterizing volatile organic compound (VOC) concentrations and estimating VOC loads was evaluated at three karst springs in Tennessee. During a 6-month period, water samples for VOC analyses were collected weekly at all three springs and as frequently as every 20 min during storms at the two springs with variable water quality conditions. Total 6-month loads for selected VOCs were calculated, and VOC data were systematically subsampled to simulate and evaluate several potential sampling strategies. Results from the study indicate that sampling strategies for karst springs need to be developed on a site-specific basis. The use of fixed sampling intervals (as infrequently as quarterly or semiannually) produced accurate concentration and load estimates at one of the springs; however, additional sampling was needed to detect storm-related changes at a second spring located in a similar hydrogeologic setting. Continuous discharge data and high-frequency or flow-controlled sampling were needed at the third spring, which had the most variable flow and water quality conditions. The lack of continuous discharge data at the third spring would substantially affect load calculations, and the use of fixed sampling intervals would affect load calculations and the ability to detect pulses of high contaminant concentrations that might exceed toxicity levels for aquatic organisms. ?? 2006 National Ground Water Association.