In the last decade Italy has experienced a consistent decline in the number of agricultural firms. Beyond the new definition of agricultural firms, this structural change of farms is characterized by exit of small farms and increasing farm size. This paper aims at analyzing the determinants of the net exit of Tuscan farms from the market during the period 2000 - 2007 both at the farm and the territorial level. The study combines data from two different sources: the 2000 census of Agriculture and three waves (2003 - 2005 - 2007) of the European “Farm and Structure Surveys” (FSS) realized in Italy by ISTAT (the National Institute of Statistics) . The resulting sample of Tuscan farms amounts approximately to 3000 agricultural firms. The exit probability of Tuscan farms from the market is estimated trough a bayesian hierarchical probit model where the group level coefficients correspond to the Local Labour Systems (LLS) i.e. a set of neighboring municipalities in which people live and work. Several variables related to farm, family, and geographical characteristics of the area are used as independent variables to investigate their net effect on the decision to exit. Results show that, among others, farm size, age of the farm operator, type of the holding have played a key role on exit. On one hand, higher farm size and professional nature of the activity lower the probability of exit. On the other hand, exit probabilities are higher for farms in which the farm operator is older nearer to retirement and without young members in his family that can replace him. Likelihood of exit is higher in areas (LLS) characterized by higher population density as the land use competition and possibly the richer labour market associated to these areas increases the exit behaviour. However given the same population density, exit probability is lower in “urban” LLS perhaps because of the proximity to remunerative market outlets for farm products.