An A to G substitution, rs925013, in the promoter of the prostate-specific antigen gene (PSA) was recently found to be associated with promoter activity and circulating PSA levels. The objective of this study was to test the associations between rs925013 and another A to G substitution, rs266882, in the PSA gene with prostate cancer risk using a population-based case-control study of 821 prostate cancer cases and 734 controls carried out in Perth and Melbourne, Australia. The study focused on young (i.e., < 70 years) and aggressive cases (i.e., well-differentiated tumors were excluded). Cases in the Melbourne arm of the study (N = 638) were followed up prospectively for an average period of 8.2 years and deaths from prostate cancer ascertained through record linkage to study the possible association between genetic variants and disease-specific survival. PSA-circulating levels were measured in controls to test the association with the genetic variants using a cross-sectional design. Linear regression of log PSA levels, unconditional logistic regression, Cox regression, and haplotype analyses were undertaken. For rs925013, the G allele was associated with an increased risk of prostate cancer [odds ratio, 1.4; 95% confidence interval (95% CI), 1.1-1.7; P = 0.001], and the hazard ratio for survival for cases homozygous for the G allele compared with cases homozygous for the A allele was increased but not statistically significant (hazard ratio, 2.3; 95% CI, 1-5.6; P = 0.06). For rs266882, there was no association with overall prostate cancer risk and survival (all P > 0.1). Men homozygous or heterozygous for the G/G (rs925013/rs266882) haplotype were at higher risk of prostate cancer than men homozygous for the A/A haplotype (odds ratio, 1.3; 95% CI, 1.1-1.7; P = 0.009). Adjusted geometric means of circulating PSA levels in controls were similar in men with zero, one, and two copies of the G allele in rs266882 (1.2, 1.1, and 1.3 ng/mL, respectively; all P > or = 0.2) and rs925013 (1.1, 1.2, and 1.5 ng/mL, respectively; all P > 0.1). For rs925013, our study provides good evidence of association with prostate cancer risk, marginal evidence of association with survival, and little evidence of detectable association with circulating PSA levels in controls. We found no evidence of an independent association between rs266882 and any of the outcomes. The genotypes and haplotypes studied might be associated with the PSA gene function or be in linkage disequilibrium with other unmeasured and functional variants in the PSA or other genes.