This research investigated whether features of examination questions influence students with dyslexia differently to others, potentially affecting whether they have a fair opportunity to show their knowledge, understanding and skills. A number of science examination questions were chosen. For some questions two slightly different versions were created. A total of 54 students considered by their teachers to have dyslexia and a matched control group of 51 students took the test under exam conditions. A dyslexia screening assessment was administered where possible and some students were interviewed. Facility values and Rasch analysis were used to compare performance between the versions of the same question and between those with and without dyslexia. Chi-square statistics found no statistically significant differences in performance between groups or between question versions. However, some tentative implications for good practice can be inferred (e.g. avoiding ambiguous pronouns, using bullet points). (Contains 1 note, 5 tables, and 4 figures.)