Common bean (Phaseolus vulgaris L.) is a traditional crop in much of Latin America, where it is often planted into soils containing numerous, sometimes ineffective, indigenous rhizobia. The presence of these indigenous organisms can limit response to inoculation. Because of this, we have sought bean cultivars that will nodulate preferentially with the inoculant strain, and have previously reported on the preference between the bean cultivar RAB39 and strains of Rhizobium tropici. We have detailed this interaction using the inoculant-quality strain UMR1899. In the present study the root tip marking (RTM) technique was used to demonstrate that this preference in nodulation was evident, even when inoculation with UMR1899 was delayed up to 8 relative to that with Rhizobium etli UMR1632. In contrast to studies with other legumes, roots of RAB39 were not predisposed to nodulate with UMR1632, even though preexposed to this strain for considerable periods of time. The presence of UMR1899 actually reduced nodulation by UMR1632 substantially, even when inoculation with UMR1899 was significantly delayed. When UMR1899 and UMR1632 were applied to separate halves of a split-root system, the number of nodules on the side receiving UMR1632 was less than for the half root inoculated with UMR1899, but the differences were not significant. This suggests that the preference response is not systemic but requires proximity between the strains involved. UMR1899 produced more than 50% of the nodules even when the ratio of UMR1632:UMR1899 in the inoculant was 10:1. The results are further evidence of a stable and marked preference of RAB39 for UMR1899, which warrants a more detailed study at the field level.Key words: Phaseolus vulgaris L., common bean, delayed inoculation, strain preference, cell proportions.