The major purposes of this study were to examine gender difference inrnmathematics achievement as a function of math self-efficacy and spatialrnvisualization and the effects of the latter two (predictor) variables on mathrnperformance.rnA math achievement test, a math self-efficacy scale, and a spatialrnvisualization test were administered to a random sample of 196 femalernand 198 male ninth grade students from four high schools of GujjirnAdministrative Zone. Then the data were analyzed using t-test, correlationrnanalysis, and Analysis of Covariance (ANCOVA).rnAnalysis of the mean differences disclosed statistically significant genderrndifferences (at .01 level) in favor of males in all the variables (math selfefficacy,rnspatial ability, and math achievement). The results of therncorrelation analysis unraveled that math self-efficacy had significantlyrnstronger relationship to math achievement than that of spatialrnvisualization. According to the stepwise multiple regression, as depictedrnby coefficients of multiple determination (R squared), the predictorrnvariables had significant joint effect on the criterion variable. Of this totalrneffect, much of the variance in performance was accounted for by genderrnfollowed by math self-efficacy. On the other hand , the results of thernanalysis of covariance showed that after statistically controlling the effectsrnof math self-efficacy and spatial visualization, gender difference in mathrnachievement was still significant. However the overall effect of gender,rnprior to controlling the two covariates, was significantly reduced.rnIt was concluded that these differential math achievements between boysrnand girls were not only due to differences in their self-efficacy and spatialrnabilities but also due to other potential factors accounting for differences inrnmathematics learning of male and female students.rnPaying attention to gender difference in math achievement and to thernfactors (cognitive and non-cognitive), that could potentially affect students'rnmathematics learning, in the teaching learning process; understandingrnstudents' self-confidence in mathematical capabil ity and taking remedialrnmeasures for extremely high or low self-efficacy beliefs; and improving thernstudents spatial visualizations through educational interventions arernamong the practical implications of the present study.