Abstract: Long life and high reliability were required for the primary pump of sodium-cooled fast reactor（SFR）, and a large number of tests were limited for the cost of R＆D and test, test cycles and test conditions and other factors. For the top bearing of the primary pump has completed 582 start-stop tests without failure, synthesis probability distribution estimation method and virtual expanding estimation method were introduced and the wear life reliability for the top bearing of the primary pump was evaluated by using the two methods. The calculation results show that the synthesis probability density function obtained by the synthesis probability distribution estimation method is closer to the true probability density function, and the virtual expanding estimation method has higher precision. The results provide a theoretical basis for the reliability estimation of the primary pump components in the case of very small samples, which has significant engineering significance in saving test expenses, reducing the times of test and shortening the research cycle. The problem of reliability assessment with only one experimental datum was solved.