Molecular analysis of recurrent tuberculosis has revealed that a second episode may be caused by a strain of Mycobacterium tuberculosis other than that involved in the first infection, thus indicating that exogenous reinfection plays a role in recurrence. We focused on two aspects of reinfection that have received little attention. First, we evaluated whether a lack of methodological refinement could lead to inaccurate assignment of mixed infections as exogenous reinfection, in which a differential selection of each of the coinfecting strains occurred over time. We used the mycobacterial interspersed repetitive-unit-variable-number tandem-repeat (MIRU-VNTR) method to genotype 122 isolates from 40 patients with recurrent tuberculosis. We identified 11/40 (27.5%) cases with genotypic differences between the isolates involved in the sequential episodes. Major genotypic differences were found in 8/11 cases, suggesting exogenous reinfection; in the remaining 3 cases, subtle genotypic differences were observed, probably indicating microevolution from a parental strain. In all cases, only a single strain was detected for the isolate(s) from each episode, thus ruling out the possibility that reinfection could correspond to undetected mixed infection. Second, we analyzed the infectivity of a selection of 12 strains from six cases with genotypically different strains between episodes. No main differences were observed in an ex vivo model of infection between the strains involved in the first episodes and those involved in the recurrent episodes. In our setting, our results suggest the following: (i) the possibility of misassignment of mixed infection as exogenous reinfection is improbable, and (ii) bacterial infectivity does not seem to play a role in exogenous reinfection.