Purpose: In this study the levels of staff turnover reported in the nursing home literature (1990-2003) are reviewed, as well as the definitions of turnover used in these prior studies. With the use of primary data collected from 354 facilities, the study addresses the various degrees of bias that result, depending on how staff turnover is defined in nursing homes. Design and Methods: Data came from a survey mailed to administrators of 526 nursing homes during March 2003. Facilities were located in four states: Missouri, Texas, Connecticut, and New Jersey. Three hundred and fifty-four responses were received, giving a response rate of 67%. Results: The 1-year turnover rates identified in these facilities were 119%, 89%, 87%, 57%, and 48%, for nurse aides, licensed practical nurses, registered nurses, administrators, and directors of nursing, respectively. However, findings show that the definition of turnover used could influence the relative difference between these reported rates by as much as 47%. Implications: As a result of measurement error, turnover rates may be misrepresented in prior studies. On the basis of the results of this investigation, several recommendations are made for future data-collection initiatives that will eliminate some of this measurement error. (Contains 4 tables.)