Examined the long-term effects of participating in a field experiment on the effects of control and predictability-enhancing interventions. 40 retirement home residents who had initially benefited from being exposed to a specific positive predictable or controllable event (visits by college students) were assessed at 3 different intervals after the study was terminated. Health and psychological status data collected 24, 30, and 42 mo after the study indicated no positive long-term effects attributable to the interventions. In fact, groups that had initially benefited from the interventions exhibited precipitous declines once the study was terminated, whereas groups that had not benefited remained stable over time. Theoretical and ethical implications are discussed.