The common way to analyse the prognostic role of selected variables in cutaneous melanoma patients is by means of Cox proportional hazard model. The prognostic effect of the simultaneous presence of more than one independent variable in the same patient is, however, difficult to establish. This hampers the possibility of tailoring a survival expectance for a selected patient as well as to communicate it to the patient himself/herself. The objectives of the study were to compare information on cutaneous melanoma prognosis from multivariate Cox proportional hazard model and from Classification And Regression Trees analysis. Classification And Regression Trees analysis is an automatic method that splits data by means of a binary recursive process creating a 'tree' of groups with different profiles according to the analysed outcome, for example, the risk of death. This approach automatically produces data that is easily interpreted by clinicians. A total of 1403 invasive cutaneous melanoma patients, 1110 from the Tuscan Cancer Registry and 293 from the Reggio Emilia Cancer Registry, Italy, were included. Cases were incident during 1996-2001 and followed up at the end of 2003. Cox proportional hazard model and Classification And Regression Trees analysis were applied to the following variables: age, sex, Breslow thickness, Clark level, registry, subsite and morphologic type. The Classification And Regression Trees analysis identified 10 categories with statistically different survival; this results were summarized into six classes of different risks based on Breslow thickness, age and sex. The best prognostic group (5-year observed survival, 98.1%) included those subjected with Breslow less than 0.94 mm and age 19-44 years. The same thickness but an older age (50-69 years) was associated with a statistically significant different prognosis (5-year observed survival, 92.8%). The Cox proportional hazard model found sex, age, Breslow thickness, Clark and morphologic type to have a significant independent prognostic value. In conclusion, compared with the conventional approach based on Cox hazard model, Classification And Regression Trees analysis produces data closer to the clinical need of defining the prognostic profile of a specific patient. This may help the clinician both in the communication of risk and in the follow-up strategy.