The cag pathogenicity island (cagPAI) has been reported to be the major virulence determinant in Helicobacter pylori-related diseases. In the present study, the diversity of the cagA gene and the integrity of the cagPAI in 158 H. pylori strains from Varanasi (North India) and Hyderabad (South India) were studied by amplifying the cagA gene (approximately 3.5 kb), followed by PCR-RFLP analysis. The results revealed significant differences in the cagA gene and the integrity of the cagPAI between North and South Indian isolates. Of 158 isolates, 40 (34.8 %) from Varanasi and 20 (46.5 %) from Hyderabad were found to carry an intact cagPAI. A partially deleted cagPAI was present in 75 (65.2 %) isolates from Varanasi and 23 (53.5 %) from Hyderabad. None of the isolates showed complete deletion of the cagPAI. Differences in the cagA 5' and 3' regions were also noted, and 11 isolates (8 from Varanasi and 3 from Hyderabad) that were cagA negative with primers for the 5' region turned out to be cagA positive with primers for the 3' variable region. It is tentatively concluded that the 3' variable region may be a better marker for cagA typing. The results also showed that the majority of the isolates harboured the Western-type EPIYA motif. PCR-RFLP analysis of the cagA gene showed 29 distinguishable digestion patterns, and cluster analysis of RFLP types from a random selection of 32 isolates placed all of the isolates into 5 groups. These results demonstrate that significant differences in the cagPAI occur among isolates from North and South India, and that RFLP of cagA could be employed for elucidating genetic variations among various isolates of H. pylori.