Objective. To perform a systematic review of 30- and 90-day mortality rates in patients undergoing hip or knee arthroplasties. Methods. Five databases were searched for English-language studies of mortality in hip or knee arthroplasties and the following data were extracted: patient characteristics (age, sex, ethnicity), arthroplasty characteristics (unilateral vs bilateral, hip vs knee), system factors (hospital volume and surgeon volume), year of study, etc. Mortality rates were compared across variable categories; proportions were compared using relative risk ratios and 95% confidence intervals. Results. Out of 650 titles and abstracts, 80 studies qualified for analysis. Of these, 35%, 34%, and 31% of studies provided 30-, 90-, and > 90-day mortality rates. Overall 30-day mortality rates across all types of arthroplasties were 0.3%; 90-day, 0.7%. For those reports with specific rates, 30-day mortality was significantly higher in men than women [1.8% vs 0.4%, respectively; relative risk (RR) 3.93, 95% CI 3.30-4.68] and in bilateral versus unilateral procedures (0.5% vs 0.3%; RR 1.6, 95% CI 1.49-1.72), but no differences were noted by the underlying diagnosis of osteoarthritis (OA) versus rheumatoid arthritis (0.4% vs 0.3%; RR 0.77, 95% CI 0.48-1.24). 90-day mortality showed nonsignificant trends favoring women, OA as the underlying diagnosis, and unilateral procedures. Conclusion. Several demographic and surgical factors were associated with higher 30-day mortality rates following knee and hip arthroplasties. More studies are needed to examine the effect of body mass index, comorbidities, and other modifiable factors, in order to identify interventions to lower mortality rates following arthroplasty procedures. The Journal of Rheumatology Copyright © 2011. All rights reserved.