The study sought to investigate the extent to which teacher made tests predict pupils performance in final O level Mathematics examinations in Seke Secondary Schools. Twoschools from Seke s five secondary schools were randomly chosen for the study. Quota samplingwas used to select four classes from the chosen two schools eight classes and systematicsampling was used to select a sample of 48 pupils from the four classes 320 pupils. Mid-yeartest and final examination results for the 48 pupils were extracted from official documents (markschedules and results slip duplicates). Interviews with four of the twelve Mathematics teachersin the Seke Secondary Schools were conducted to elicit information on factors that maycompromise predictability of pupil performance at final examinations through use of teachermadetest scores. Descriptive survey research design was used in the research study.Collected data were analysed using Pearson Product Movement Correlation, Paired T-Test andOne-Way ANOVA..