Serotyping of clinical isolates is a widely used technique for epidemiologic study of group B streptococcal infections. However, serotyping cannot definitively determine epidemiologically related or unrelated isolates. We investigated the use of restriction endonuclease analysis (REA) with both conventional agarose gel electrophoresis (AGE) and pulsed-field gel electrophoresis (PFGE) in 50 isolates of the major serotypes of group B streptococci. Single digestion with HindIII and HaeIII and double digestion with HindIII and then EcoRI were used for conventional AGE, and digestion with SmaI was used for PFGE. The molecular profile of one strain was compared with those of the strains within the same serotype as well as with the profiles from strains of different serotypes. Among 10 type Ia, Ia/α, Ia/α+β, and Ia/R1 isolates and depending on the restriction enzyme used, we found between five and six REA patterns by conventional AGE and seven by PFGE; among 4 type Ib/α+β isolates we found 2 to 4 REA patterns by conventional AGE and 4 by PFGE; among 21 type II, II/α, II/β, II/α+β, and II/R4 isolates, we found 11 REA patterns by both AGE and PFGE; and among 14 type III, III/R1, and III/R4 isolates, we found from 7 to 12 different REA patterns by AGE and 10 by PFGE. In total, among 13 serotypes and one nontypeable strain, we found 29 to 31 REA patterns by conventional AGE and 33 by PFGE. A particular REA pattern found within a serotype was different from the patterns found in the other serotypes, suggesting that REA analysis by using conventional AGE or PFGE is a sensitive method for analyzing genetic relatedness and diversity in group B streptococci and has potential value in molecular epidemiologic studies.