BACKGROUND: There is an on-going debate about whether to perform surgery on early stage localised prostate cancer and risk the common long term side effects such as urinary incontinence and erectile dysfunction. Alternatively these patients could be closely monitored and treated only in case of disease progression (active surveillance). The aim of this paper is to develop a decision-analytic model comparing the cost-utility of active surveillance (AS) and radical prostatectomy (PE) for a cohort of 65 year old men with newly diagnosed low risk prostate cancer. METHODS: A Markov model comparing PE and AS over a lifetime horizon was programmed in TreeAge from a societal perspective. Comparative disease specific mortality was obtained from the Scandinavian Prostate Cancer Group trial. Direct costs were identified via national treatment guidelines and expert interviews covering in-patient, out-patient, medication, aids and remedies as well as out of pocket payments. Utility values were used as factor weights for age specific quality of life values of the German population. Uncertainty was assessed deterministically and probabilistically. RESULTS: With quality adjustment, AS was the dominant strategy compared with initial treatment. In the base case, it was associated with an additional 0.04 quality adjusted life years (7.60 QALYs vs. 7.56 QALYs) and a cost reduction of [euro sign]6,883 per patient (2011 prices). Considering only life-years gained, PE was more effective with an incremental cost-effectiveness ratio of [euro sign]96,420/life year gained. Sensitivity analysis showed that the probability of developing metastases under AS, utility weights under AS are a major sources of uncertainty. A Monte Carlo simulation revealed that AS was more likely to be cost-effective even under very high willingness to pay thresholds. CONCLUSION: AS is likely to be a cost-saving treatment strategy for some patients with early stage localised prostate cancer. However, cost-effectiveness is dependent on patients' valuation of health states. Better predictability of tumour progression and modified reimbursement practice would support widespread use of AS in the context of the German health care system. More research is necessary in order to reliably quantify the health benefits compared with initial treatment and account for patient preferences.