Abstract: With the deepening of population aging in China, the pricing method of long-term care insurance for the elderly has become a hot issue in actuarial direction. Based on the data of CLHLS from 2014 to 2018, the health statuses of the elderly are further divided into six states on the basis of the traditional three and four state Markov model. It uses the Markov model to calculate the value of each state. The transition strength matrix and transition probability matrix of health state are solved by Robinson power function, which takes gender and age into account. Then, the premium years of 65, 75 and 85 years old are estimated by using double random Lee-Carter model, random walk model and life expectancy formula, which provides a theoretical reference for the pricing of long-term care insurance in China.