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. The purpose of this report was to determine how much change in standard instrument scores represents a clinically detectable improvement or deterioration. The Veterans Affairs (VA) Cooperative Study of Clozapine in Refractory Schizophrenia evaluated 423 patients on clozapine or haloperidol. Symptoms and quality of life scales were completed at baseline; 6 weeks; and 3, 6, and 12 months. Among patients judged as "improved" by clinicians, the average percentage changes were a 21 percent decrease in Positive and Negative Syndrome Scale (PANSS) scores and a 26 percent increase in Quality of Life Scale (QLS) scores across all followup periods. The change in mean seven-point item scores were -0.46 (PANSS) and 0.23 (QLS). A major gain in clinically assessed improvement to "much better" was associated with a 45 percent decline in PANSS scores and 50 percent increase in QLS scores (change in mean seven-point item scores -0.88 and 0.92, respectively). Thus, modest changes in psychometric scales assessing symptoms and quality of life reflect clinically detectable improvement.