To investigate health gain and its predictors during inpatient and associated day patient treatment. METHOD: Consecutive admissions to two inpatient units for children and young adolescents in northwest England were studied (N = 55). Ascertainments were made from multiple perspectives, including family, teacher, clinician, and an independent researcher. Measures were taken at referral, admission, discharge, and 6-month follow-up; health gain was inferred from change scores on measures. Recruitment lasted from late 1995 to 1997; follow-up was completed during 1998. Independent variables tested as predictors included assessments of presenting symptoms, therapeutic alliance, and family functioning. RESULTS: Significant health gain during hospitalization was found on most measures and sustained to follow-up. There was no symptom change during the waiting-list control condition. Health gain was predicted independently by child and parental therapeutic alliance with the unit early in hospitalization and by preadmission family functioning. Externalizing problems did well if accompanied by good alliance. CONCLUSIONS: Assessment of health gain from multiple perspectives is possible and valuable. Inpatient treatment has significant therapeutic effect. Predictors for health gain lie in process variables of therapeutic alliance and family functioning rather than presenting symptoms. The results are discussed in relation to clinical practice and future research.