Various estimation models and numerous sampling methods were used to estimate the size of a pumpkinseed sunfish (<i>Lepomis gibbosus</i>) population in a 1.16 hectare Virginia farm pond. After estimates were obtained, the pond was drained in order to compare estimated population size with actual population size. In all cases, population estimates were significantly less than actual population size. An examination of the underlying assumptions indicated that biases were attributed to mortality of marked fish, an increased catchability of marked fish due to marking, and the ability of fish to learn to avoid capture gear. Different gear combinations and use of different estimation models had no significant effect on the accuracy of estimates obtained. The study was designed to employ commonly used methods of estimating population size in a situation where underlying assumptions appeared relatively valid. It is apparent from the results of this study that underlying assumptions are often not met to the degree necessary to provide useful data. / Master of Science