OBJECTIVE To compare clinical features, treatments and outcomes in men with non-metastatic prostate cancer (PCa) according to whether they were referred for symptoms or elevated prostate-specific antigen (PSA) level. PATIENTS AND METHODS This study used data from the South Australia Prostate Cancer Clinical Outcomes Collaborative database; a multi-institutional clinical registry covering both the public and private sectors. We included all non-metastatic cases from 1998 to 2013 referred for urinary/prostatic symptoms or elevated PSA level. Multivariate Poisson regression was used to identify characteristics associated with symptomatic presentation and compare treatments according to reason for referral. Outcomes (i.e. overall survival, PCa-specific survival, metastasis-free survival and disease-free survival) were compared using multivariate Cox proportional hazards and competing risk regression. RESULTS Our analytical cohort consisted of 4 841 men with localized PCa. Symptomatic men had lower-risk disease (incidence ratio [IR] 0.70, 95% confidence interval [CI] 0.61-0.81 for high vs low risk), fewer radical prostatectomies (IR 0.64, CI: 0.56-0.75) and less radiotherapy (IR 0.86, CI: 0.77-0.96) than men presenting with elevated PSA level. All-cause mortality (hazard ratio [HR] 1.31, CI: 1.16-1.47), disease-specific mortality (HR 1.42, CI: 1.13-1.77) and risk of metastases (HR 1.36, CI: 1.13-1.64) were higher for men presenting with symptoms, after adjustment for other clinical characteristics; however, risk of disease progression did not differ (HR 0.90, CI: 0.74-1.07) amongst those treated curatively. Subgroup analyses indicated poorer PCa survival for symptomatic referral among men undergoing radical prostatectomy (HR 3.4, CI: 1.3-8.8), those aged >70 years (HR 1.4, CI: 1.0-1.8), men receiving private treatment (HR 2.1, CI: 1.3-3.3), those diagnosed via biopsy (HR 1.3, CI: 1.0-1.7) and those diagnosed before 2006 (HR 1.6, CI: 1.2-2.7). CONCLUSION Our results suggest that symptomatic presentation may be an independent negative prognostic indicator for PCa survival. More complete assessment of disease grade and extent, more definitive treatment and increased post-treatment monitoring among symptomatic cases may improve outcomes. Further research to determine any pathophysiological basis for poor outcomes in symptomatic men is warranted.