Among the hypotheses offered to explain the conflicting results of various trials of BCG vaccination, the one invoking gross differences in the immunogenic properties of the various BCG vaccines (strains) has received most attention. Indeed, for many years research in BCG vaccination has been directed almost entirely towards improving the quality of vaccines. the anticipated benefit of this work, however, is not borne out by the results of the latest trial in India. Whereas it cannot be excluded that technological advances may have resulted in the worst vaccines ever, it also remains possible that the basic hypothesis was less relevant than presumed. the quality of the vaccine may be of little importance and a different explanation should be sought for the observed differences in protection. One possible lead is the observation that BCG vaccination consistently appeared to be of poor efficacy under conditions where, in a vaccination programme, even an effective vaccine would have little impact on the tuberculosis problem, i.e. where the majority of cases originated from the population already infected. The hypothesis offered is that in such populations the pathogenesis of tuberculosis is different from that prevailing under the clearly exceptional conditions under which BCG vaccination was effective. Rather than being the direct result of primary infection, tuberculous disease (observed) may be the result of reinfection. In this case BCG vaccination cannot be expected to have a protective effect.