Normally, selection of students for higher education is based on previous performance. Whether or not previous performance is a good predictor varies for different examining bodies. This study evaluates the predictive ability of Tanzania’s Ordinary Certificate of Secondary Education (OCSE) performance to Advanced Certificate of Secondary Education (ACSE) performance using linear regression and chi-square tests. By linking OCSE of 2003 and 2004 with ACSE of 2006 and 2007, respectively, research results indicate that the OCSE is a good predictor for ACSE. The predictive ability of OCSE improves as data are disaggregated based on whether or not students studied at the same school and based on subject combinations, especially for science-based subjects. This indicates that admission criteria, generally employed for ACSE, serve the academic purposes they are assumed to be serving.