Aim-To compare restriction fragment length polymorphism (RFLP) and random amplified polymorphic DNA (RAPD) methods for the epidemiological typing of Mycobacterium tuberculosis.Methods-Thirty one M tuberculosis cultures originating from patients in the Canton of Berne in Switzerland, which had previously been typed by RFLP, were subjected to RAPD analysis. Cultures were coded so that the investigators were blind to the RFLP results until RAPD analysis was complete.Results-The 31 cultures of M tuberculosis were divided into nine groups by RFLP and eight groups by RAPD. Generally there was good correlation between the groups identified by the two techniques, with the exception of strains that had only one copy of IS6110. Both methods subdivided isolates that were placed in a single group by the other method.Conclusions-RAPD analysis is quick, simple, and useful for the comparison of small numbers of isolates. RFLP is more reproducible and therefore better suited for the accumulation of RFLP fingerprints for long term local surveillance and large epidemiological studies.