Abstract: Objective To understand the underreporting of all death cause mortality in Pengzhou, Sichuan province, from 2012 to 2013. Methods The all death cause mortality data among the local residents during this period were collected from local death cause surveillance system, township family planning offices and village committees. in Pengzhou the data from different sources were matched with the names, name's phonetic letters, sex, date of birth and living address of death cases. Log-liner model was used for fitting the data from 3 sources to estimate the all death cause mortality. Results By using capture and recapture strategy, it was estimated the all death cause mortality was 13 594 (95% CI: 13 537-13 651), and the mortality rate was 1.77%, higher than that in death cause registry (1.41%) and the average of samples (1.68%). By using the mortality estimated by capture and recapture strategy as denominator, the underreporting rate of average samples was 5.29%; the underreporting rate of death cause surveillance was 20.37%, higher than the rate (15.92%) by using average samples as denominator. Conclusion Any samples of death are incomplete, even three samples were used, the underreporting still exists. Capture recapture method can be used to correct the error caused by underreporting death by death surveillance system, and the number of death can be estimated more accurately.