Increasing the number of plants per beat bucket sample significantly reduced the number of samples necessary to estimate the mean density of common predators at study sites in Arizona and Texas. Ten plants per sample unit required the least number of total samples, but required the most sampling time. Preliminary analyses indicated that for most predator groups, a sample unit of 3 plants per bucket resulted in the most precise estimate of density at the lowest cost as measured in time to collect the samples. Estimates of the most efficient sample unit size for Texas and Arizona were similar for predator groups common to both locations with the exception of those for crab spiders.