We investigated the effect of sampling site access on estimates of fish abundance in 2 eastern Oklahoma streams. Centrarchid species were sampled by electrofishing at public and remote access sites on Baron Fork Creek in northeastern Oklahoma and Glover River in southeastern Oklahoma. We verified differences in recreational use and habitat between access types in both streams. Recreational use was generally higher at public than remote access areas in each stream. Public areas in Glover River had higher fish densities, were deeper, and had more instream cover than remote areas. However, mean density of fish, mean depth, and frequency of cover types at public and remote areas in Baron Fork Creek were not significantly different. Although we did not observe a consistent trend in fish abundance between streams at public and remote access sites, our findings indicate that accessibility sampling from public access areas may yield biased estimates of population size. Therefore, we urge caution when making inferences about populations based on samples taken solely from these areas.