ObjectivesTo quantify the preference of patients with diabetes mellitus (DM) for primary healthcare (PHC) institutions in China to redirect the patient flow and improve health outcomes.DesignCross-sectional study. Discrete choice experiment (DCE) surveys asked patients with DM to choose between hypothetical institutions that differed in the medical service capacity, out-of-pocket (OOP) medical costs per month, travel time, the attitude of medical staff and the availability of diabetes drugs.SettingShandong province, China.ParticipantsThe participants were 887 patients with DM from 36 urban communities and 36 rural villages in Shandong province. One participant did not provide any DCE answers and a further 57 patients failed the internal consistency test. 829 fully completed surveys were included in the final data analysis.Main outcomes and measuresA mixed logit model was used to calculate the willingness to pay and predict choice probabilities for PHC institution attributes. Preference heterogeneity was also investigated.ResultsAll five attributes were associated with the preferences of patients with DM. The OOP medical costs and the medical service capacity were the most influential attributes. Improvements simultaneously in the attitude of medical staff, drug availability and travel time increased the likelihood of a patient’s PHC institution choice. Preferences differed by region, annual household income and duration of diabetes.ConclusionsOur patient preference data may help policymakers improve health services and increase acceptance of choosing PHC institutions. The OOP medical costs and medical service capacity should be regarded as a priority in decision-making.