Background: The objective of this study is to explore the nature of the association of discrete systems of human resource management practices and the rate of voluntary turnover of registered nurses (RNs) in a large sample of Canadian acute care hospitals and nursing homes. Healthcare quality, cost and access are impacted by excessive turnover of nursing staff. The rate of annual turnover for registered nurses has been estimated to be about 20 percent in Canada, with a total cost of almost $27,000 for each RN. Healthcare organizations that employ large numbers of registered nurses are keen to learn more about the role that organizational policies and practices play in reducing voluntary turnover. The relationship of human resource management (HRM) practices and employee turnover has generally been under-investigated. Three HRM systems, composed of bundles of discrete workplace or employment practices, were conceptualized in order to assess the relationship of HRM practice systems with the rate of RN turnover, at the establishment or facility level. Three HRM practice systems were constructed to test their systemic impact on nursing turnover: a technical HRM system that focuses on bureaucratic policies and practices that regulate the employee-employer relationship; a quality-of-worklife HRM system that includes a variety of employee-centered and family-friendly employment practices; and a high-involvement HRM system that utilize workplace arrangements that increase the commitment, engagement, accountability and participation of nurses. Methods and analysis: The chief nursing officers of 2,208 Canadian hospitals and long-term care facilities received the study questionnaire, generating responses from a total of 705 establishments. Each of the three HRM practice systems is estimated at each facility and comprised eight (8) discrete employment or workplace practices. Using estimates of voluntary RN turnover as the dependent variable, a step-wise ordinary least squares (OLS) regression was performed on each HRM practice system. Results: After controlling for facility characteristics and for local labour market conditions, the quality-of-worklife HRM employment system (p < .05) and high-involvement HRM work system (p < .05) are found to be related with lower voluntary turnover, yet their overall relationship with facility RN turnover appears to be quite modest. Our technical HRM practice system is not found to be associated with facility RN turnover. Conclusion: HRM systems comprising employee-friendly employment practices and high-involvement work practices may play an important (but modest) role in reducing RN turnover.