Single-species flow-through toxicity tests were conducted to determine the times-to-death of two indigenous fish to South Florida--least killifish (Heterandria formosa) and mosquitofish (Gambusia affinis)--from acute exposure to endosulfan sulfate. Mortalities were recorded within 8-h periods from test initiation to termination at 96 h. The 96-h LC(50)s for least killifish and mosquitofish estimated using the trimmed-Spearman-Karber method were 2.0 and 2.3 microg/l, respectively. An accelerated failure time model was used to estimate times to death at selected concentrations. Data were fit to log-normal, log-logistic, and Weibull distributions. Acute toxicity data fit to the Weibull distribution produced a better relative fit than log-normal or log-logistic distributions for both toxicity tests. The survival-time profiles and associated statistics illustrate the benefit of considering exposure duration as well as concentration when predicting acute risk to species' populations. Both toxicity tests had similar outcomes from exposure to endosulfan sulfate, with least killifish being slightly more likely to die at lower concentrations and shorter time periods than mosquitofish. From the models generated by the toxicity tests, times-to-death for least killifish and mosquitofish were estimated for environmentally relevant concentrations of total endosulfan at a site of concern in South Florida. When the results from the current toxicity tests were compared to environmental concentrations from previous screening-level ecological risk assessments, the durations necessary to potentially kill 10% or more of the populations of the two native south Florida fish species were estimated to be 77 and 96 h for least killifish and mosquitofish, respectively. However, the exposure values included the alpha and beta isomers as well as endosulfan sulfate; therefore, an understanding of their toxicity might be important in understanding the survival dynamics of fish species in endosulfan sulfate-contaminated sites.