Objective: To assess differences in serial mean utility scores, from Health Utilities Index (HUI) Mark 3 (HUI3) measurements for patients undergoing treatment for acute lymphoblastic leukemia (ALL) in childhood. These scores will be used for identifying determinants of health. Methods: Patients were aged 5 years or older at time of health-related quality of life (HRQL) survey and enrolled in the Dana Farber Cancer Institute 95-001 clinical trial. Parents-of-patients completed self-administered HUI questionnaires during the four major phases of treatment: induction of remission (t1); CNS prophylaxis (t2); intensification (t3); and maintenance (t4). HRQL scores are on a scale where 0.00 equals dead and 1.00 equals perfect health. Significant differences between treatment phases, for mean HRQL and single-attribute utility scores, were assessed using 1-way ANOVA and paired t-test. Magnitude of change in mean HRQL scores between treatment phases was assessed by effect size. Differences in mean HRQL scores of >0.03 are clinically important. Results: A total of 375 patients were surveyed (55.2% males). Patients with complete sets of parental measurements at all treatment phases were included in this analysis (n=86). Mean HRQL score at t1=0.68 (SD=0.306), t2=0.74 (SD=0.245), t3=0.77 (SD=0.267), and t4=0.88 (SD=0.204) (p<0.001). There was substantial inter-patient variability in HRQL scores within treatment phases. The effect sizes were 0.200 for t2-t1, 0.120 for t3-t2 and 0.395 for t4-t3. Mean HRQL change scores were 0.06 for t1 to t2 (p=0.085), 0.03 for t2 to t3 (p=0.416), and 0.10 for t3 to t4 (p=0.001). For single-attribute scores significant differences between treatment phases were observed for ambulation (p<0.001), emotion (p<0.001), pain (p<0.001). Conclusion: Mean HRQL, ambulation, emotion and pain scores generally improved over time. The large inter-patient variability may reflect, in part, the considerable heterogeneity of treatment-related side effects among patients. Further analyses will explore whether demographic or diagnostic risk factors contribute to this variability.