Data from 2 aerial surveys of waterfowl using catfish ponds were used to compare sampling efficiences between 2 stratified sampling designs and 2 completely random designs. Completely random designs produced slightly lower coefficients of variation (CV) than post-stratified sampling designs. The unbiased estimator was simplest to compute. Based on a curve established from proportions of catfish ponds sampled and CV's, sampling intensity may be reduced if lower levels of precision were acceptable. Sampling intensity can be reduced by employing a stratified design and optimal allocation of the sample size.