A binomial sampling plan for pest management of the stable fly, Stomoxys calcitrans (L.), was developed. Counts of stable flies on front legs of the same animal were independent and each leg from the same animal was considered a sample unit. The relationship between the mean number of flies per leg and the variance was determined and did not vary among farms. The relationship between the mean number of flies per leg and the proportion of legs with zero, one or less, and two or less flies (P0, P1, and P2) was determined and used as the basis of the binomial sampling plan. Predicted values of the mean number of flies per leg from P0, P1, and P2 were close to observed values of the mean number of flies per leg. Equations are presented for calculating the variance of a predicted value of the mean number of flies per leg using values of P0, P1 and P2 determined by sampling. Operating characteristic (OC) curves are also presented for determining the probability of making a treatment decision error at an economic threshold of one fly per leg (P0 = 0.47) using binomial sampling or direct counting. OC curves for binomial sampling with n = 50 legs were close to those for direct counting with n = 10 legs. Recommendations concerning the use of binomial sampling for stable flies are presented.