Systematic follow-up of patients after initial treatment of cancer is stressful with heavy cost. The aim of such follow-up is to detect asymptomatic lesions in viewing to increase global survival and quality of life. For an individual patient, the issue of cancer recurrence is a binary event. However, when developing surveillance strategies for large groups of patients, knowledge of the risks (tumor biology, natural history of the disease), the benefits (potential efficacy of salvage therapy) and diagnosis test performances is necessary to formulate a rationale and resource effective follow-up algorithm. The Bayes'nomogram is useful to assess diagnosis test. Soft tissue sarcoma is a example of such demonstration.