Four randomly selected nursing groups were assigned to three experimental groups and one control group to test the relative impact of three experimental nursing schedules, using a before-after design. The three experimental treatments were (1) straight shifts; (2) regular schedule but with unlimited requests for changes; and (3) individual station-designed schedules. Before treatment, score differences between the experimental and control groups were limited to one of 36 highly reliable scales specifically constructed and pretested to gauge effects of scheduling. This single difference was judged not to be significantly related to experimental outcomes. Because of a poor job market situation, retention was not affected significantly by any of the three treatments, but root causes of turnover were. Results of the experiment showed that individual station-designed schedules triggered the most changes that favor retention. In contrast, the other two treatments unexpectedly increased nurses' own sense of marketability and reduced teamwork among nurses. Reasons accounting for the results are discussed in the text.