Background: Both ability (measured by power tests) and non-ability (measured by preference tests) individual difference measures predict academic school outcomes. These include fluid as well as crystalized intelligence, personality traits, and learning styles. This paper examines the incremental validity of five psychometric tests and the sex and age of pupils to predict their General Certificate in Secondary Education (GCSE) test results. Aims: The aim was to determine how much variance ability and non-ability tests can account for in predicting specific GCSE exam scores. Sample: The sample comprised 212 British schoolchildren. Of these, 123 were females. Their mean age was 15.8 years (SD 0.98 years). Method: Pupils completed three self-report tests: the Neuroticism-Extroversion-Openness-Five-Factor Inventory (NEO-FFI) which measures the "Big Five" personality traits, (Costa & McCrae, 1992); the Typical Intellectual Engagement Scale (Goff & Ackerman, 1992) and a measure of learning style, the Study Process Questionnaire (SPQ; Biggs, 1987). They also completed two ability tests: the Wonderlic Personnel Test (Wonderlic, 1992) a short measure of general intelligence and the General Knowledge Test (Irving, Cammock, & Lynn, 2001) a measure of crystallized intelligence. Six months later they took their (10th grade) GCSE exams comprising four "core" compulsory exams as well as a number of specific elective subjects. Results: Correlational analysis suggested that intelligence was the best predictors of school results. Preference test measures accounted for relatively little variance. Regressions indicated that over 50% of the variance in school exams for English (Literature and Language) and Maths and Science combined could be accounted for by these individual difference factors. Conclusions: Data from less than an hour's worth of testing pupils could predict school exam results 6 months later. These tests could, therefore, be used to reliably inform important decisions about how pupils are taught.