The reliability of an assessment is defined as the extent to which candidates' results would remain stable if the entire assessment exercise was repeated. Whilst numerous studies have evaluated the reliability of written examinations, relatively little has been done to quantify the reliability of internal teacher assessment within schools. This is unfortunate since several high-stakes qualifications in England depend upon the reliability of internal assessment, often in the form of teacher-marked coursework (or similar). The use of coursework is justified in terms of ensuring the validity and authenticity of the learning experience. However, quantitative evidence of reliability is often lacking. This paper will attempt to infer something about the overall reliability of coursework by comparing its value to that of written examinations taken at the same time in predicting future examination scores. This study will use several years of assessment data from General Certificate of Secondary Education (GCSE) and General Certificate of Education Advanced level (GCE A level) assessments in History and English Literature in England. The results will show that coursework is often just as predictive as externally marked tests in forecasting future performance. Since reliability is a necessary pre-condition for (predictive) validity, this suggests that it may be more reliable than is often recognised.