Recent efforts to identify non-cognitive predictors of academic achievement and school success have largely focused on self-constructs such as self-efficacy, self-concept and anxiety that are measured with respect to a specific domain (e.g. mathematics). We extend the measurement of the non-cognitive realm in education to incorporate both social and psychological adjustment variables and ratings of confidence in addition to these self-constructs. Our findings show that confidence explains most of the variance in achievement captured by the other self-constructs combined, and that psychological adjustment variables add little to the equation. Furthermore, in contrast to other cognitive and non-cognitive variables, confidence accounts for 46.3% of total variance in achievement, while measures of previous cognitive performance in combination with other non-cognitive variables account for 40.5% of the total variance. We discuss the ways in which confidence is important in education.