The purpose of this study was to investigate the predictive validity of competence testrnadministered by Center of Competence (COC). The study was conducted on 220 (170rnCompetent and 50 Not yet competent) candidates of COC from the year 2000 to 2002.rnThe candidates for the study were selected by availability sampling. Supervisory ratingrnforms (Criterion 1) were used to collect the necessary data of the participants andrnAppraisal questiormaire (criterion 2) were use to triangulate the result of Criterion 1.rnQuantitative research methodology was used to analyze the data. The result of thernquestionnaire (Criterion 2) revealed that the competent participants have performancernwhich exceeds the expectation set for their position while the not yet competentrnparticipants have performance meet the expectation set for the position and the result ofrnt-test revealed that there is a significant performance difference between the competentrnand not yet competent candidates. The result of correlation analysis revealed that therncompetence test administered by COC moderately predicts on job performances of thernTVET graduates. The result of one way ANOV A also showed statistically significantrnmean difference exists between the selected sectors of TVET. The result of independentrnt-test revealed that there is statistically significant difference between the mean ofrncompetent and not yet competent participants and the result of chi-square showed thatrnthere is relationship between on job performance and COC scores of the candidates.rnFinally it was concluded that the competence test predicted the performance ofrncandidates of the test. And the test can be generalized to the selected sectors of TVETrnthat the competence test provided. The competent candidates have better performancernthan that of the not yet competent candidates. Further more that it would be fine ifrnemployers and other stakeholders of COC use the result of COC to make decision