The study was conducted among 156 grade 5 students (78 girls and 78 boys)rnand 157 grade 9 students (77 girls and 80 boys) of the schools of DebrernBirhan town. The main objective of the study was to examine thernrelationships of previous math experience, specific math self-efficacy, generalrnmath self-efficacYi and subsequent math achievement among students ofrngrades 5 and 9. The major statistics used to achieve this end were chisquare,rnmultiple linear regression and t-test. One of the principal findingsrngave a hint that goes with Bandura's (1977) contention that previousrnexperience is the most influential source of self-efficacy development. At bothrngrade levels, task specific math self-efficacy predicted subsequent mathrnachievement well. Besides, previous math experience made weak butrnsignificant contribution in predicting subsequent math achievement among 9thrngraders, but not among 5th graders. However, sex and general math selfefficacyrnmeasures did not predict subsequent math achievement in eitherrngrade. Moreover, grade 5 students were found more confident than 9thrngraders.rnThe mean scores of boys and girls on all of the four variables wererncomparable among 51h graders. Boys and girls of grade 9 produced similarrnscores in previous math experience, specific math self-efficacy, and generalrnmath self-efficacy, but boys surpassed girls in subsequent math achievement.rnThe gender gap observed in math achievement at high school level is arnphenomenon repeatedly reported in many studies. Thus, high school teachersrnshould do whatever they can to promote females' math achievement so as torndevelop self-efficacy in math