OBJECTIVES: The main objective of this study was to analyze different approaches and methods to explore potential causal associations among prevalent pathologies. We have focused on Diabetes Mellitus (DM) and its well known association with incident neoplasia. METHODS: For this retrospective cohort study data were obtained from CSD LPD, a General Practitioner’s longitudinal database. We have evaluated the risk of neoplasia incidence among people with Diabetes mellitus compared with those without this pathology, in patients who had no reported history of benign or malign neoplasia at the start of the follow-up on January 2006. For the DM group, patients with at least one diagnosis of DM from January 2005 to December 2005 have been selected, while for the DM free group, patients without a diagnosis of DM and a date of registration in the GPs office before January 2006 have been selected. Both groups have been followed-up for 48 months. RESULTS: During the selection period 45.121 (3.4%) patients with a diagnosis of DM (Females: 22.330, Males: 22.791) and 1.290.597 (96,6%) patients  without a diagnosis of DM (Females: 690.462, Males: 600.135) have been selected.  During the follow-up 6.648 and 80.880 incident cases of neoplasms have been documented from the DM and DM free groups respectively. The mean follow-up duration was 43 and 45 months for the DM and the DM free groups respectively. CONCLUSIONS: The selected cohort has shown to match quite well with general population in terms of gender and age. The estimated prevalence of Diabetes also matches with the one of the general population. Statistical analysis has shown an Adjusted (for age and sex) Hazard Ratio of 1,88 (95% Cl 1,83-1,93) suggesting an association between DM and incident neoplasms, evidencing that GPs longitudinal databases could be a valid instrument for evaluating causal association among prevalent pathologies.