The National Senior Certificate (NSC) examinations were written for the second time in 2009 amid much criticism. In this study, scripts of candidates who wrote the NSC Mathematics examinations (papers 1 and 2) in 2009 were used as data to analyse the marks scored and then polytomous Rasch analysis was conducted for all the sub-questions to determine the level of difficulty of the questions. The purpose of applying Rasch measurement models is to explore the extent to which a test or an examination and its associated data set permit the interpretation of an underlying linear scale of ability against which to interpret overall performance and item difficulty. In the NSC data, some questions discriminated well at the lower-ability levels of candidates, but no questions were found to discriminate among higher-ability candidates.