Studies were conducted to evaluate the survival and persistence of Sinorhizobium meliloti 104A14 and two acid phosphatase-negative mutants in Kirkland (fine, mixed, thermic Udertic Paleustolls) silt loam soils with various fertility levels, and to assess the impact of inoculation on nodule occupancy and soil microbial community structure in the inoculated alfalfa (Medicago sativa L.) rhizosphere. Recovery of the inoculated strains was 100% (in the order of 10 8 cellsg - 1 soil) immediately following inoculation to soils, but decreased from 10 8 cellsg - 1 soil to undetectable levels in a nutrient-poor soil within 32 days. In a nutrient-rich soil, approximately 2-3% (4.7-7.43x10 6 cellsg - 1 soil) of the mutants and 23% (5.84x10 7 cellsg - 1 soil) of the wild-type inocula persisted for more than 64 days. Survivability and persistence of the wild-type S. meliloti were significantly greater than that of the genetically modified acid phosphatase negative mutants in all the soils tested. The persistence and nodule occupancy of the introduced S. meliloti in sterile and non-sterile soils were also tested for two repeated alfalfa growth periods in the same plant growth units, with a 1 month interval in between and no additional inoculation for the second period. Nodule occupancy of the introduced S. meliloti in non-sterile soils ranged from 30 to 60% for the first period and 85 to 100% for the second period. Our results suggest that survival and persistence of S. meliloti was enhanced by alfalfa cultivation and increased soil fertility, but impaired by mutation of acid phosphatase genes regardless of phosphorus nutritional levels. Moreover, inoculation with genetically modified S. meliloti strain 104A14 promoted indigenous bacterial growth in soil (increased bacterial population from 1.4x10 6 to 4.3x10 6 cellsg - 1 soil), but not the growth of fungi and yeast. However, inoculation of the wild-type S. meliloti or genetically modified mutants did not result in significant changes in microbial community structure as indicated by EP indices and ratios of r/K strategists.