Intact soil-core microcosms were used to compare persistence of Pseudomonas chlororaphis 3732RN-L11 in fallow soil and on wheat roots with field releases at diverse sites. Parallel field and microcosm releases at four sites in 1996 were repeated with addition of one site in 1997. Microcosms were obtained fresh and maintained at 60% soil water holding capacity in a growth chamber at 70% relative humidity, a 12-hour photoperiod, and constant temperature. Persistence of 3732RN-L11 was measured at each site in field plots and microcosms at 7-21 day intervals, and in duplicate microcosms sampled at an independent laboratory. Linear regression slopes of field plot and microcosm persistence were compared for each site, and between identical microcosms sampled at different sites, using log10 transformed plate counts. Microcosm persistence closely matched field plots for wheat roots, but persistence in fallow soil differed significantly in several instances where persistence in field plots was lower than in microcosms. Analysis of weather variations at each site indicated that rainfall events of 30-40 mm caused decreased persistence in fallow soil. Cooler temperatures enhanced persistence in field plots at later time points. Inter-laboratory comparison of regression slopes showed good agreement for data generated at different sites, though in two instances, longer sampling periods at one site caused significant differences between the sites. Soil characteristics were compared and it was found that fertility, namely the carbon to nitrogen ratio, and the presence of expanding clays, were related to persistence. These microcosm protocols produced reliable data at low cost, and were useable for pre-release risk analyses for microorganisms.