Determining the fate and survival of genetically-engineered microorganisms released into the environment requires the development and application of accurate and practical methods of detection and enumeration. Several experiments were performed to examine quantitative recovery methods that are commonly used or that have potential applications. In these experiments, Erwinia herbicola and Enterobacter cloacae were applied in greenhouses to Blue Lake bush beans (Phaseolus vulgaris) and Cayuse oats (Avena sativa). Sampling indicated that the variance in bacterial counts among leaves increased over time and that this increase caused an overestimation of the mean population size by bulk leaf samples relative to single leaf samples. An increase in the number of leaves in a bulk sample, above a minimum number, did not significantly reduce the variance between samples. Experiments evaluating recovery methods demonstrated that recovery of bacteria from leaves was significantly better with stomacher blending, than with blending, sonication, or washing and that the recovery efficiency was constant over a range of sample inoculum densities. Delayed processing of leaf samples, by storage in a freezer, did not significantly lower survival and recovery of microorganisms when storage was short term and leaves were not stored in buffer. The drop plate technique for enumeration of bacteria did not significantly differ from the spread plate method. Results of these sampling, recovery, and enumerations experiments indicate a need for increased development and standardization of methods used by researchers as there are significant differences among, and also important limitations to, some of the methods used. (Copyright (c) 1991 American Society for Microbiology.)