OBJECTIVES: To estimate transition probabilities between cervical cancer and its precancerous lesions for understating the natural history of cervical cancer among Chinese women. METHODS: A 30-cycle Markov model was established to simulate the natural history of cervical cancer among Chinese women among 30-59 years old. Initial model parameters, including initial proportion of each health state, death probabilities, and transition probabilities, were estimated based on cervical screening studies, literature review and expert consultation in China. We adjusted transition probabilities in the model to fit age-specific prevalence of each health state with empirical data from government-sponsored cervical cancer screening programs for rural women in 1999-2008. In addition, we compared model estimates with age-specific annual incidence synthesized from screening research programs by the Chinese Academy of Medical Sciences Cancer Institute in rural China. RESULTS: Annual transition probabilities for cervical intraepithelial neoplasia (CIN) 1 to CIN2, CIN2 to CIN3/carcinoma in situ (CIS), and early cervical cancer to invasive cancer were 0.145, 0.120 and 0.180, respectively. In addition, age-specific progression probabilities were 0.065-0.092 for healthy to HPV infection, 0.070-0.120 for HPV infection to CIN1, and 0.020-0.035 for CIN3/CIS to early cervical cancer. As for regression, transition probabilities for CIN1 to healthy/ HPV infection, CIN2 to CIN1 were 0.253 and 0.230, respectively. Age-specific probabilities for HPV infection to healthy, CIN3/CIS to CIN2 were 0.430-0.490 and 0.030-0.080, respectively. Based on these transition probabilities, model predictions for cervical cancer were largely in accordance with findings from population-based screening programs and were within 99% confidence intervals of cervical cancer incidence data from research programs. CONCLUSIONS: Transition probabilities for cervical cancer and its precancerous lesions estimated from the Markov model are reliable. Our model based on these transition probabilities is useful for understanding the natural history of cervical cancer.