Abstract: The prior distribution of the mean time between failures (MTBF) of a machining center was obtained according to the previous failure data from similar machine tools produced in batches, and the compatibility of the prior information with the in-situ data was examined using the Wilcoxon-Mann-Whitney rank sum test. The reliable life of the machining center was predicted online based on the Bayesian theory combined with the prior distribution data and the small sample failure data of the machining center. The results show that the failure time distribution of the machining center follows the two-parameter Weibull distribution, i. e., β=1.9093, α=919.4951. The point estimation value of the MTBF is 815.80 (h) and its interval estimation value is [576.70 (h), 1232.42 (h)]. The method proposed is suitable for the small sample situation, which is meaningful for the reliabe life prediction of special manufacturing equipment.