16 psychiatric patients were observed in and were asked to describe their reactions to 6 psychiatric ward subsettings. Findings indicate: (1) that persons, settings, and Person * Setting interactions accounted for statistically significant and important proportions of the total variance, both in responses to questionnaires and in actual behavior; and (2) that these proportions remained remarkably stable over a 3-mo interval, except for a consistent tendency for the proportions attributable to settings to increase. Implications discussed include that different psychiatric ward subsettings may be differentially therapeutic to different Ss, and that the findings bear on the upper limit of validity indexes obtainable with standard interview or rating predictions of S behavior. (17 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1969 American Psychological Association.