Abstract: This research has been rooted in the Bandura’s (1986) Social Cognitive Theory of learning. The paper investigated how teacher mentorship relates to teachers’ job performance when moderated by head teachers’ management styles. From a pragmatic realist viewpoint, a cross-sectional survey design with concurrent mixed method approach was used in data collection, analysis and interpretation. A questionnaire comprising of Teacher mentorship experience scale, Teacher job performance scale and Leadership behaviour scale was filled by 286 teachers while four school administrators were interviewed. The results revealed a moderate level of mentorship (M = 35.95, SD = 6.78, Min =19, Max=50) and a moderate level of teachers’ job performance (M = 41.58, SD = 5.04, Min=13, Max = 50). Nonetheless, there was a positive significant correlation (r = .398, p < .001) between mentorship and teacher job performance. A multiple regression analysis using Process plugin in SPSS generated a significant moderation effect of head teacher leadership styles on the link between teacher mentorship and teacher job performance within the Johnson-Neyman region of significance. Autocratic leadership increased the effect of mentorship on job performance (B= -0.02, p =.05) by head teachers more than any other styles. This implies that head teachers have to be more directing and instructing in order for their teachers to perform their duties effectively.