The increasing demand for public health services requires the implementation of efficient management practices to ensure the quality and sustainability of the system. Since the 2000s, governments have been contracting with the third sector for the provision of health services, arguing that this would lead to efficiency gains. This study aimed to analyze and compare the efficiency of primary health care provided directly by public administration and by health social organizations (HSOs), private entities from the third sector, in municipalities in southeastern Brazil. To this end, a two-stage output-oriented Data Envelopment Analysis (DEA) BCC model was used, along with statistical procedures such as jackstraw and boxplots to identify outliers and a Tobit regression model. The analyses revealed that primary health care coverage and municipal development in education and health are positively associated with efficiency in all municipalities analyzed, and that total personnel expenses in relation to the total health budget and per capita health expenditures were also significant factors affecting efficiency in cities with more than one hundred thousand inhabitants. The results indicated that the average efficiency scores were low for both management models, and in small municipalities, direct administration obtained better results, contrary to the official discourse that delegating service provision to HSOs would be more efficient. In summary, and in light of the study, we suggest that the HSO model is still fragile as an organizational alternative within the scope of primary health care in the Brazilian Unified Health System (SUS), as there is no robust evidence of efficiency or effectiveness gains that justify the transfer of public resources to the private sector for the execution of this policy.