Problems associated with quantitative estimations of mixed populations of several species of predators on apple trees are discussed. The problems arise largely because of the inherently high variability in such populations and because of the mechanical difficulties of counting predators on some basic unit of tree. In this investigation two devices used in sampling are described and their efficiency rated. One, a vacuum sampler, that removed predators from samples of leaf clusters on the tree failed to meet the standards set. The other, called the shaker sampler, tumbled clusters about in a revolving wire screen cage to dislodge predators for counting. The amount of processing with this device to remove practically all predacious specimens was calibrated. When the shaker-sampler technique was applied to studies in two orchards it proved to be practical for quantitative estimation of predator populations per large limb. The error associated with estimations of the density of individual species was usually very high so that pooling of the results for several or all species was required to reduce the variance to an acceptable level.