Service utilization measures from the psychiatric case registers for urban South-Verona and rural Portogruaro in North East Italy for the period 1983–9 were used to identify associations with socio-demographic variables from the 1981 census in schizophrenia and related disorders as well as in all diagnoses. The patterns of service use were broadly similar, except that Portogruaro has significantly more community contacts, and has about twice the treated incidence and prevalence of schizophrenia. The census data showed that unmarried and unemployed people were more likely to live alone in the urban than in the rural area. In South-Verona the most strongly associated predictor variables, both for schizophrenia and all diagnoses, are: living alone, unemployment, percentage of the total population who are dependents and the percentage who are divorced, separated or widowed. In contrast, in Portogruaro there were no consistent associations between census and service use variables. Stepwise multiple regression models using three census predictor variables accounted for over 85% of the variance in South-Verona utilization rates. The results indicate that the strongly predictive associations previously described in England hold in urban South-Verona, but not in rural Portogruaro, and may be related to the effect of cities in clustering seriously disabled psychiatric patients in areas of low-cost housing where they live in relative social isolation.