BACKGROUND Hip and knee arthroplasty infections are associated with considerable healthcare costs. The merits of reducing the postoperative surveillance period from 1 year to 90 days have been debated. OBJECTIVES To report the first pan-Canadian hip and knee periprosthetic joint infection (PJI) rates and to describe the implications of a shorter (90-day) postoperative surveillance period. METHODS Prospective surveillance for infection following hip and knee arthroplasty was conducted by hospitals participating in the Canadian Nosocomial Infection Surveillance Program (CNISP) using standard surveillance definitions. RESULTS Overall hip and knee PJI rates were 1.64 and 1.52 per 100 procedures, respectively. Deep incisional and organ-space hip and knee PJI rates were 0.96 and 0.71, respectively. In total, 93% of hip PJIs and 92% of knee PJIs were identified within 90 days, with a median time to detection of 21 days. However, 11%-16% of deep incisional and organ-space infections were not detected within 90 days. This rate was reduced to 3%-4% at 180 days post procedure. Anaerobic and polymicrobial infections had the shortest median time from procedure to detection (17 and 18 days, respectively) compared with infections due to other microorganisms, including Staphylococcus aureus. CONCLUSIONS PJI rates were similar to those reported elsewhere, although differences in national surveillance systems limit direct comparisons. Our results suggest that a postoperative surveillance period of 90 days will detect the majority of PJIs; however, up to 16% of deep incisional and organ-space infections may be missed. Extending the surveillance period to 180 days could allow for a better estimate of disease burden. Infect Control Hosp Epidemiol 2017;38:147-153.