To cope with the rising price of drugs, in 2009 the Chinese government launched a large pharmaceutical reform. Its key element is the implementation of a National Essential Medicine List, leading to a reorientation of incentives for health services providers. Health facilities are not anymore allowed to make any profit on drug sales (“zero mark-up policy”), while this used to be their main source of financing. Different compensation schemes have been implemented by the authorities. In a context of refunding of the financing structure of health care facilities, it is crucial to understand how the reform has affected –or not- health care facilities activity and efficiency. This study relies on a survey data from a sample of 30 Township Hospitals of the rural prefecture of Weifang (Shandong province). Using a two-stage procedure, it aims at assessing the THs’ technical efficiency scores and then at identifying the determinants of this efficiency. The first stage is realized with a non-parametric frontier approach, the so-called ‘partial frontier’ method, order-m to deal with the problem of dimensionality. The identification of the determinants of efficiency requires panel data models, with random individual effects. Results show that the average efficiency remains constant between 2006-09 and 2010-12, around 0.65. The most significant and robust factors of technical efficiency are the share of subsidies in the TH incomes for the first sub-period (negative effect), and the number of covered inhabitants per bed (positive effect). It suggests that drug reform hasn’t improve primary health facilities efficiency, certainly because the reform did not tackle with success the issues of the financial barriers to universal access to healthcare -out-of-pocket payments from patients-, and of the perception of quality of public healthcare.