The persistence of rotenone was compared between a cement-lined pond (0.04 hectare) and an earthen-bottom pond (0.02 hectare) treated with 5 I?L Noxfish/L (250 I?g rotenone/L) during spring, summer, and fall. Water temperatures on the days of treatment in each season were 8, 22, and 15A?C, respectively. Both ponds were filled with pond water from a common source 1 week before each of the three treatments. Water samples (filtered and unfiltered) and sediment samples were analyzed by high-performance liquid chromatography to monitor the decrease of rotenone until residues were at or below the detection limit (<2.0 I?g/L for water and < 25 ng/g for sediments). The loss of rotenone from water generally followed a first-order rate ofdecay. Rotenone disappeared two to three times faster in the earthen pond than in the concrete pond. The rotenone half-life times in the spring, summer, and fall treatments were 3.7, 1.3, and 5.2 d, respectively, in the concrete pond, and 1.8, 0.7, and 1.8 d in the earthen pond. Rates of decay in both ponds were directly correlated with water temperature. Filtered water samples from both ponds contained less rotenone than unfiltered water, indicating that some rotenone was bound to suspended material. The highest concentration of rotenone in sediment samples was 102 ng/g; residues decreased to below the detection limit within 14 d in the spring treatment and within 3 d in the summer and fall treatments.