_Background and objective:
The shortage of health professionals generates inequalities in access to healthcare and is also related to misdistribution of medical specialists, multiplicity of jobs and mobility among health regions in Brazil. The aim of this study is to describe patterns of physicians’ mobility and the workforce dependence to identify mediator factors. Methods: The study design is a survey based on the National Registry of Healthcare Establishment (CNES), in which health services and professionals are registered. It was analyzed 19 specialties of physicians’ registrations in CNES in June 2018 in 435 health regions. It was performed a factorial analysis including 38 variables, which indicated two synthesis variables: medical specialists supply and exclusive physicians working in health region (72% of variance). It was performed a cluster analysis that identified different groups among health regions.

Results:
It was identified 5 clusters of health regions that correlate to mobility and dependence. In cluster 1 there are all metropolitan areas and greater cities, with all medical specialties and a high percentage of exclusive physicians (N=92 health regions). In cluster 2, there are regional hubs, with all medical specialties, with mixed exclusive and non-exclusive physicians (N= 145). In cluster 3, there are smaller and midsized cities with shortage of medical specialties with high rate of non-exclusive doctors, with high dependence from cluster 1 regions (N=119). In cluster 4, there are smaller cities with shortage of medical specialties, with mixed frequencies of exclusive and non-exclusive physicians (N=35), and higher rates of primary care physicians. In cluster 5, there is the lowest rate of specialists with greater predominance of primary care and exclusive physicians (N=44).

Conclusion:
Different patterns of mobility were identified, related to specific healthcare structure and socioeconomic characteristics. Those findings contribute to guide directed human resources redistributive strategies for each cluster of health regions.