In this paper we analyse all General Practitioners (GP) prescriptions in a Friuli-Venezia Giulia area (North-eastern Italy). The sample included of 181 GPs and 242,564 patients with 1,191,122 prescribed items. The regional Health Data Base is the source of all data. Data analysis was performed according to the multiple regression and LISREL models. We assessed for all GPs the years of medical profession, patients listed, percent of patients exent from prescription rates, patients over 60, number of active compounds prescribed and related drug companies. Two factors, the number of active compounds prescribed and patients listed, significantly contributed to the multiple regression model. The model explained 56% of variation in prescribing pointing out the importance of those factors in influencing the cost per patient. LISREL model shows a causal chain going from the years of medical profession to cost per patient passing through the number of active compounds and drug companies related. In fact, the older the GP the less the cost per patient and number of active compounds used. A main finding is that the cost per patient is highly influenced by the number of different active compounds prescribed. As in other studies the percent of patients exent from prescriptions rates and aged over 60 are somehow fundamental in influencing the cost per patient. Finally we discussed models and literature on the issue.