OBJECTIVE: Health related quality of life (HRQOL) has been shown to be an important predictor of survival in patients with cancer. The aim of this study is to evaluate the role of serial measurements of HRQOL in predicting survival for newly diagnosed localized prostate cancer patients. METHODS: Data from CaPSURE, a longitudinal disease registry of men with prostate cancer, was used to identify a study population. HRQOL were measured by SF-36 and were assessed bi-annually by mailed questionnaires. All-cause mortality was reported by treating physician or next-of-kin and confirmed by death certificates. To account for changes in HRQOL over time Cox proportional hazards regression with time-dependent covariates analysis was conducted. Univariate Cox proportional hazards model was fit for each HRQOL domain to determine which were related to survival time. We then fit a multivariate model using the HRQOL domains found to be significantly related to survival time and also controlled for baseline clinical risk and age at diagnosis. Categorical variables were created for HRQOL based on the baseline distribution of the lower 10th percentile and the reminder of the patients. RESULTS: A total of 5610 patients met the study criteria and 322 (5.6 %) patients had died. Mean survival time (defined as time from treatment to either death or last follow up) was 33.3 months. Lower scores on Physical Function, Role Physical, Role Emotional, Vitality, Mental Health, Social Function, Bodily Pain and General Health were strongly associated with death (HR between 2.9 to 5.6). In multivariate analyses Physical Function, Role Physical and General Health remained significantly associated with survival (HR=2.0, 95% CI=1.4-2.8; HR=1.9, 95% CI=1.3-2.6; HR=1.9, 95% CI=1.4-2.5 respectively). CONCLUSIONS: Lower scores on serial measurements of physical and general health domains appear to be significant predictor of survival even when controlling for known clinical factors.