Previous studies have described an increased risk of malignancy in subjects diagnosed with rheumatic conditions, most notably rheumatoid arthritis (RA). Our aim was to quantify and compare risks for site-specific malignancy among hospitalized patients with RA, osteoarthritis (OA) and other rheumatic conditions in a nationwide, population-based cohort. Subjects were identified from Scottish hospital in-patient records from 1981 to 1996 and followed up by computer linkage of the Scottish Cancer Registry and the national registry of deaths. Expected cancer incidence was calculated from national cancer rates and related to the observed incidence by the standardized incidence ratio (SIR). Among RA patients, there was an increased risk for hematopoietic [males SIR= 2.13, 95% confidence interval (CI) 1.7-2.7; females SIR = 1.76, 95% CI 1.5-2.1], lung (males SIR = 1.32, 95% CI 1.2-1.5; females SIR = 1.44, 95% CI 1.3-1.6) and prostate (SIR = 1.26, 95% CI 1.0-1.6) cancers. Reduced risk were seen for colorectal cancer (males SIR = 0.87, 95% CI 0.7-1.1; females SIR = 0.71, 95% CI 0.6-0.9) and, among females, stomach cancer (SIR = 0.70, 95% CI 0.5-1.0). The excess risk for hematopoietic cancer and the reduced risk for colorectal and stomach cancers were sustained over 10 years of follow-up. An overall decreased risk of cancer was observed for patients with OA; the greatest reductions were observed for colorectal (males SIR = 0.88, 95% CI 0.8-1.0; females SIR = 0.84, 95% CI 0.8-0.9), stomach (males SIR = 0.79, 95% CI 0.7-0.9; females SIR = 0.66, 95% CI 0.6-0.8) and lung (males SIR = 0.72, 95% CI 0.7-0.8; females SIR = 0.84, 95% CI 0.8-0.9) malignancies, with decreased risks generally still evident at 10 years of follow-up. Our results support several previous findings regarding the incidence of hematopoietic and colorectal malignancies in RA patients. In addition, we have shown a large decrease in stomach cancer among patients with OA and females with RA that warrants further investigation since it may provide clues to possible prevention strategies. To further our knowledge about the underlying mechanisms of altered risk in cancer patients with rheumatic conditions, population studies requiring primary data collection are required.