Instrument-based scores are often used as outcome measures. However, little is known about what changes in scores mean in terms of a clinical assessment of improvement or deterioration. OBJECTIVE: The purpose of this study was to determine how much change in standard symptom and HRQOL instrument scores represents a clinically detectable improvement or deterioration for schizophrenia patients. METHODS: The VA Cooperative Study of Clozapine in Refractory Schizophrenia evaluated 423 patients with clozapine or haloperidol. Symptoms (PANSS) and quality of life (QLS) scales were competed at baseline, 1.5, 3, 6, and 12 months. Changes were calculated as percentage improvement and as changes in a mean 7-point item scale based on clinicians ratings of patient status compared to baseline. RESULTS: Based on clinician- assessed improvements from baseline, PANSS scores improved by 21%; QLS scores by 26% when patients were rated as improved, and 45%, 50%, respectively when rated as much better. Mean 7-point item scores improved by 0.46 points PANSS, 0.23 points QLS for improved patients, and by 0.88, 0.92, respectively for much better patients. Minimal changes were seen for all items when patients were rated as unchanged. The relationship between clinician ratings and score changes was stable over multiple assessments in this large sample. CONCLUSION: These data describe the amount of change that represents clinically important improvement related to treatment based on commonly used symptom and QOL scales. Planning sample sizes for clinical trials can be based on percentage change in scores or mean item scores to aid clinicians and researchers in understanding meaningful treatment effects.