To assess factors contributing to variation in trawl data, eight stations in Whitewater Bay, Everglades National Park, Florida were sampled each month by roller-frame trawling from December 1968 to November 1969. A total of 52,190 fishes representing 66 species was captured. Analyses of catch data for 15 of these species indicated that the number of individuals per sample conformed to the Poisson distribution at low sample densities and to the negative binomial as population size increased and aggregation or “contagion” became evident. Tests of selected methods of data transformation indicated that a logarithmic scale transformation with the negative binomial statistic k was often effective in rendering such data amenable to parametric statistical testing. Sampling procedures and tidal stage had no direct effect on catch rates, although some evidence was found for an interaction between them. Vegetation density appeared to be the most important factor determining the differences observed between stations.