A bioassay relating number of Agrobacterium tumefaciens cells in the inoculum quantitatively to the number of crown‐gall tumors initiated on primary pinto bean leaves is described. Variability in estimation of infectious titers by this assay is similar to that observed in comparable plant virus assays, most determinations showing standard errors of 20% of the mean tumor per leaf value. The assay has the advantages of speed and practicality. The efficiency of the system is low, typically requiring between 105 and 106 bacteria for each tumor initiated. Infectivity titers of 103‐104, however, are readily obtained from stationary phase cultures. Statistical analysis of the infectivity titration curve indicates that a single bacterium is the usual infectious unit. The assay is specific within the family Rhizobiaceae to the species Agrobacterium tumefaciens and Agrobacterium rubi. A. tumefaciens strains IIBNV6 and ATCC # 11095 were non‐infectious, while strain B6 was the most infectious of the strains tested. The infectivity of the latter strain is shown to decrease about 4‐fold between early log and stationary phases of growth. Changes in the growth medium or in the dilution‐inoculation medium failed to alter the infectivity of the bacterium.