ABSTRACT: BACKGROUND:China adopted family doctor (FD) to help achieve "Healthy China 2030" through providing continuous, comprehensive, and life-cycle contract services. However, there is a disparity between actual and targeted FD use, as residents continue to visit specialists in large hospitals. The government implemented initiatives to improve residents' willingness to sign up with and visit their FDs. Factors that influence contract behavior are therefore significant for frontier policy research. METHODS:Two survey waves were conducted in Shanghai (2013 and 2016). The first wave included 2754 people and the second 1995 people. Exploratory factor analysis was used to synthesize "satisfaction" as a predictor of contract behavior. Pearson's chi-square, pooled and logistic regression models were used to estimate associations between influencing factors and contract behavior, and clarify variations in factors across the two waves. RESULTS:Four factors were extracted from 15 satisfaction items: "Treatment Environment," "Medical Technology," "Service Specification" and "Service Attitude". Consistent with descriptive analysis, longitudinal analysis showed sociodemographic characteristics (age, education, marital status, and hukou) were significant predictors of contract behavior. The odds ratio of non-communicable diseases (NCD) patients for contract behavior was 2.218 times that of residents without NCD. Contract behavior was positively correlated with awareness of FD services (OR = 21.674, 95%CI = 15.043-31.229), satisfaction with Service Attitude (OR = 1.210, 95%CI = 1.009-1.451), and visit compliance (OR = 1.959, 95%CI = 1.564-2.452). Over time, the odds ratios of the married, Shanghai hukou, NCD, and awareness of FD services declined from 0.456, 1.795, 2.492, 28.690 to 0.443, 1.678, 1.910 and 14.031 respectively, while those of age, and visit compliance increased from 1.027, 1.521 to 1.041 and 2.305 respectively. In 2016, an education-contract gradient had formed (the higher the education level, the higher probability of signing with a FD), whereas high school education had the highest odds ratio (OR = 1.163,95%CI = 0.740-1.827) in 2013. Service Attitude was the only significant satisfaction-related predictor (OR = 1.358, 95%CI = 1.001-1.842) in 2016, compared with "Treatment Environment" (OR = 1.224, 95%CI = 1.001-1.496) and "Service Specification" in 2013(OR = 1.270, 95%CI = 1.040-1.552). CONCLUSIONS:Except for the socio-demographic variables, NCD, awareness of FD services, satisfaction and visit compliance were significant predictors of contract behavior with FDs. The effect of visit compliance had increased over time while NCD and awareness of FD services were losing impact over time. Significant satisfaction factors had also changed from "Treatment Environment" and "Service Specification" to "Service Attitude".