Human Lyme borreliosis (LB) is the most prevalent arthropod-borne infection in temperate climate zones around the world and is caused by Borrelia spirochetes. We have identified 10 variable-number tandem repeat (VNTR) loci present within the genome of Borrelia burgdorferi and subsequently developed a multiple-locus VNTR analysis (MLVA) typing system for this disease agent. We report here the successful application of MLVA for strain discrimination among a group of 41 globally diverse Borrelia isolates including B. burgdorferi, B. afzelii, and B. garinii. PCR assays displayed diversity at these loci, with total allele numbers ranging from two to nine and Nei's diversity (D) values ranging from 0.10 to 0.87. The average D value was 0.53 across all VNTR loci. A clear correlation exists between the repeat copy number and the D value (r = 0.62) or the number of alleles (r = 0.93) observed across diverse strains. Cluster analysis by the unweighted pair-group method with arithmetic means resolved the 30 observed unique Borrelia genotypes into five distinct groups. B. burgdorferi, B. afzelii, and B. garinii clustered into distinct affiliations, consistent with current 16S rRNA phylogeny studies. Genetic similarity and diversity suggest that B. afzelii and B. garinii are close relatives and were perhaps recently derived from B. burgdorferi. MLVA provides both phylogenetic relationships and additional resolution to discriminate among strains of Borrelia species. This new level of strain identification and discrimination will allow more detailed epidemiological and phylogenetic analysis in future studies.