Three field inoculation experiments, two in Florida and one in New Mexico, were conducted with Azospirillum brasilense Cd. Each of the Florida experiments evaluated two crop species. One species in each of the Florida experiments responded to inoculation with a significant dry matter yield increases of 11 to 24% and nitrogen yield increases of 9 to 39%. No inoculation response was noted in the New Mexico experiment. The responding species were Sorghum bicolor (L.) Moench (sorghum) and the interspecific hybrid between Pennisetum americanum (L.) K. Schum. (pearl millet) and P. purpureum Schumach. (napiergrass). Nonresponding species were pearl millet (Florida) and Sorghum sudanense (Piper) Staph. (New Mexico). Acetylene reduction activity of inoculated plots in Florida was low, showing no increase over the natural uninoculated background rates and, in one case, was negatively correlated with yield. Acetylene reduction activity was not measured in New Mexico. In Florida, A. brasilense populations were found to decline from 5 × 103 to 5 × 102 bacteria g of soil−1 in about 3 weeks (quadratic regressions). Continued decline to less than 102 by week 5 indicated that the inoculated bacteria did not become established in the soil in high numbers. The A. brasilense population declined at about the same rate in the New Mexico experiment. The erractic inoculation responses in these experiments are similar to those observed in earlier work at the University of Florida. The lack of acetylene reduction activity response to inoculation and the rapid population decline of the inoculated bacteria suggest that N2 fixation is not the major mechanism causing yield responses after inoculation.