The ethnic variation in the GGN and CAG microsatellites of the androgen receptor (AR) gene suggests their role in the substantial racial difference in prostate cancer risk. Hence, we performed a case-control study to assess whether GGN repeats independently or in combination with CAG repeats were associated with prostate cancer risk in South Indian men. The repeat lengths of the AR gene determined by Gene scan analysis, revealed that men with GGN repeats <or=21 had no significant risk compared to those with >21 repeats (OR 0.91 at 95% CI-0.52-1.58). However, when CAG repeats of our earlier study was combined with the GGN repeat data, the cases exhibited significantly higher frequency of the haplotypes CAG <or=19/GGN <or=21 (OR-5.2 at 95% CI-2.17-12.48, P < 0.001) and CAG <or=19/GGN > 21(OR-6.9 at 95%CI-2.85-17.01, P < 0.001) compared to the controls. No significant association was observed between GGN repeats and prostate-specific antigen levels and the age at diagnosis. Although a trend of short GGN repeats length in high-grade was observed, it was not significant (P = 0.09). Overall, our data reveals that specific GGN/CAG haplotypes (CAG <or=19/GGN <or=21 and CAG <or=19/GGN > 21) of AR gene increase the risk of prostate cancer and thus could serve as susceptibility marker for prostate cancer in South Indian men.