Based on Reliability Generalization (RG) approach, this meta-analysis study focused on 85 English papers chosen from 782 papers (in both English and Chinese) about 44-Item Big Five Inventory (BFI-44) published in the past 24 years (1991~2014). This study analyzed the amount of variability in Cronbach's coefficient alpha of every dimension in BFI-44 and examined predictor variables, including sample size, subjects’ nationality, test language, average age and gender ratio, which were used for further investigation of their impact on the reliability of BFI-44 by using regression analysis. The study results showed that the Cronbach's coefficient alpha of each dimension reach nearly 0.8. The sample size, nationality and gender ratio variables influence BFI-44 dimensions on different levels. There are two suggestions for the use of the BFI-44. First, researchers should report the Cronbach's coefficient alpha of his/her own samples to avoid the reliability induction error;secondly, researchers should pay attention to the effect of sample size, nationality and gender ratio on Cronbach's coefficient alpha.