In a large (N= 1744) study of previously hospitalized psychiatric patients, multiple follow-up attempts were made to contact the ex-patients over a 1-year period after their discharges. When contacted they were asked to provide information about their posthospital adjustment; 59.5% of the sample was reached at least once and usable data obtained either in a telephone interview or from a mailed survey form. The contacted and noncontacted people represented very different subpopulations, both demographically and in terms of typical psychiatric descriptors. Those who were of lower socioeconomic status, male, unmarried, racial minorities, and those with records of substance abuse or assaultiveness, and who were generally more severely impaired during the baseline hospitalization wereunderrepresented in the contacted group. Possible reasons for these sample biases, the implications for hospitals conducting outcome assessments (i.e.,for research and program evaluation purposes), and strategies for dealing with this kind of methodological problem are discussed.