The effect of inoculant formulation on the population dynamics of rhizobia in the pea rhizosphere was investigated using a streptomycin-resistant mutant of Rhizobium leguminosarum bv. viceae NITRAGIN128C56G (128C56G strR). The isolate was formulated into liquid, peat powder, and granular peat carriers, and was tested on pea at field sites near Saskatoon, Saskatchewan, and Beaverlodge, Alberta, in 1996 and 1997. The liquid and peat powder formulations were applied to seed while the granular inoculant was applied to soil. In three out of four site years, population dynamics were similar among formulations: an initial decline or lag period lasting 2-5 days followed by an increase to approximately 10(5) colony-forming units (CFU)/seedling by 14-28 days after planting (DAP) and, where sampled, a continuing increase from 10(7) to 10(8) CFU/plant at 63 DAP. In these same site years, nodule number (not determined at Beaverlodge in 1997) and nodule occupancy at 60 days were not significantly different among formulations. In contrast, soil populations of 128C56G strR from the liquid formulation declined to near zero by 28 DAP at Beaverlodge in 1996, when soil moisture was excessive in spring because of high rainfall. Populations increased in this treatment after this time, but remained significantly lower than the populations of the other two formulations throughout the sampling period. Pea seed yields were not significantly different among treatments in either year at Beaverlodge, but were significantly higher with granular inoculant than the noninoculated control in Saskatoon. Within inoculated treatments at Saskatoon, there were no significant differences in grain yield.