Counts of grasshoppers from Wyoming and New Mexico were related to the proportion of O.1−m2samples that contained grasshoppers through binomial sampling models. Three models were investigated. One was based on the Poisson probability density function and two were based on the Nachman and logit binomial regressions. The Nachman and logit regressions were tested extensively for an influence of predominant instar, sampling year, and location on the estimated parameters. Results indicated that instar and sampling year did not influence the parameter estimates, but location did. All models were then tested to determine which provided the smallest error for predicting densities from incidence and vice versa. Results showed that no differences occurred between models. Based on this finding, the Poisson sampling model was selected for use in wide-scale surveys because it is simple and easy to use. An independent data set from Montana was used to validate this model. Implications of this and other findings in context of using binomial sampling as a management and research tool are discussed.