Two sampling designs were compared to evaluate Oklahoma's standardized sampling procedures for electrofishing largemouth bass (Micropterus salmoides) in reservoirs. Historical subjectively-chosen fixed sites were sampled along with random sites stratified by habitat category (good, fair, and poor) at four reservoirs in central Oklahoma. The stratified categories were determined by a composite of the shape/structure of the bottom, substrate type, and type of cover available in 0.5-km transects. Using the stratified random design, three of the four reservoirs showed significantly different (P ≤ 0.05) catch per unit effort (CPUE, fish h- 1) between the habitat categories. Good habitat at those three lakes exhibited the highest CPUE while poor habitat was the lowest. CPUE was significantly lower (P ≤ 0.05) at the stratified random sites than the fixed sites at two of the four reservoirs. Furthermore, Kolmogorov-Smirnov tests and visual observations indicated length frequencies between the two sampling designs were similar. Although CPUE results were mixed between the two designs, the stratified random sampling design was recommended for largemouth bass electrofishing. This design strengthens standardized sampling procedures over the fixed site design by meeting the assumption of random samples in probability statistics which allows comparisons among reservoirs and years to be made. It also likely provides more representative estimates of the population without jeopardizing precision when compared to the fixed site design.