Most studies regarding agglomeration and productivity are focused on service sector as a whole and on manufacturing sector. Few papers have paid attention to knowledge-intensive business services (KIBS), which are mainly concerned with providing knowledge-intensive inputs to the business processes of other organizations. Therefore, the purpose of this research is to analyse agglomeration economies and productivity of knowledge-intensive business services (KIBS) located in Italy and operating in the period of 2009-2018, based on firm-level data drawn from AIDA, a commercial database collected by Bureau Van Dijk. The firm level data are merged with different variables at province and regional level drawn fromISTAT, OECD, Italian trade and marks, and Eurostat databases. We propose an empirical framework based on the characteristics of KIBS and of the local context interms of agglomeration economies to study the channels through which KIBS firms benefit from specialization and diversity externalities. The empirical evidence are also disaggregated by different KIBS subsectors to detect the heterogeneity betweenProfessional KIBS (P-KIBS) and Technological KIBS (T-KIBS), the most dynamic branches of KIBS. We include commonly used control variables for firm characteristics such as tangible and intangible assets, age, measured as the number of years since a firm’s founding data, size, measured as the total number of employees, leverage, i.e the ratio of debt to assets measured with total liabilities divided by total assets, research and development investment. We also control for regional and province level variables related to innovation, logistic access and human capital stock. Unlike manufacturing and traditional services, KIBS are characterized by relying heavily on highly skilled employment, intense interaction with clients, and professional knowledge. Hence, having access to a suitable labor force, reducing transportation and transaction costs, and increasing knowledge flows are the main channels through which agglomeration economies contribute to KIBS performance. Since estimates of agglomeration economies suffer from two main biases: unobserved heterogeneity and simultaneity, we also address those two problems through a fixed effects approach first to solvef or unobserved heterogeneity. Then we use an instrumental variables approach to investigate the direction and magnitude of potential biasarising from the endogeneity of density.