The circulating level of insulin-like growth factor-binding protein-3 (IGFBP-3) is inversely associated with the risk of prostate cancer (PCa) and its progression and may be modulated by the A/C polymorphism at position -202 in the promoter region of IGFBP-3. This study was conducted to evaluate the role of the A/C polymorphism as a genetic modifier in the etiology of PCa and its disease status. The polymorphism was analyzed by a PCR restriction fragment-length polymorphism technique in 307 PCa patients, 221 benign prostatic hyperplasia (BPH) patients, and 227 male controls. No significant difference in the genotype frequency was found between the PCa or BPH patients and controls (PCa versus control, P = 0.316; BPH versus control, P = 0.964). Regarding the tumor stage, the C allele was more frequently observed in patients having tumors with higher stage (P for trend = 0.002). When the PCa patients with localized disease (stage A + B + C) were considered as reference, those with CC and AC genotype had a significantly increased risk of metastatic disease (stage D) compared with those with AA genotype [age-adjusted odds ratio (aOR) = 3.89, 95% confidence interval (CI) = 1.42-10.64, P = 0.008, and aOR = 1.68, 95% CI = 1.01-2.79, P = 0.044, respectively]. The presence of the C allele appeared to be associated with an increased risk of metastatic PCa with a gene dosage effect (aOR = 1.82, 95% CI = 1.23-2.68, P = 0.002). Similarly, significant findings were also observed when PCa patients were compared between those with organ-confined disease (stage A + B) and those with extra-prostatic extension (stage C + D). Furthermore, the C allele was present more frequently in patients with higher tumor grade. In conclusion, the IGFBP-3 -202 A/C polymorphism was not associated with susceptibility to PCa and BPH in Japanese men, but the presence of the C allele may cumulatively increase the risk for tumor metastasis and for having tumors with a biologically more aggressive phenotype. Because of the significant differences in incidence of clinically evident PCa according to racial backgrounds, the conjecture should be further examined in different racial populations.