As tourism is becoming one of the most important sources of economic growth at the local level, it is imperative to identify and assess the relevant determinants of tourism flows. This paper investigates this issue by carrying out an econometric analysis based on the origin-destination (OD) spatial interaction models, which fully account for the spatial dependence generally featured by tourism flows. We contribute to the current debate by analyzing the tourism flows for the complete set of 107 Italian provinces (11449 OD flows) in terms of 2009 arrivals. Besides geographical distance, the explanatory variables include both pull and push locations’ characteristics to assess their relative role in determining the distinctive traits of emissiveness and attractiveness for all the provinces. We thus consider income, density, accessibility (low-cost flights, transport infrastructure), a set of cultural (museums) and natural (park areas, coasts, well-preserved beaches) factors and other amenities (renowned restaurants). The main results point out that there is a great deal of spatial correlation induced by neighboring provinces at both origin and destination, which is systematically overlooked if one relies only on the gravity specification. Once one controls for such a complex kind of dependence, most of the explanatory variables exhibit the expected effect, with distance and population density showing a negative impact on tourists’ decisions when choosing a specific destination, while amenities, accessibility and income turn out to be effective determinants of tourism flows.