This article uses an empirical contribution to better understand the project management office (PMO). PMOs are an important aspect of project management practice. Their design and management is complicated by the great variability found among PMOs in different organizations. Lack of consensus on their structure and the roles they undertake prevent the establishment of formal standards on PMOs. Having a typology of PMOs can make the great variability much more manageable. However, the typology should be grounded in reality. The aim of this article is to exploit a rich database of descriptions of 500 PMOs to identify patterns in the data that can form the bases for one or more typologies of PMOs. Data on both the organizational context and the characteristics of PMOs were explored. The search for the bases of a typology relies on the identification of statistical associations (1) between the characteristics of PMOs and characteristics of their organizational context, (2) between the different characteristics of PMOs themselves, and (3) between the performance of PMOs and the characteristics of both PMOs and their organizational context. The analysis explores each of these avenues successively in the search for characteristics that are good or poor candidates for forming the basis of a typology of PMOs. The results of the analysis are then integrated into a model.