This research set out to compare the quality, length and nature of (1) exam responses in combined question and answer booklets, with (2) responses in separate answer booklets in order to inform choices about response format. Combined booklets are thought to support candidates by giving more information on what is expected of them. Anecdotal evidence suggests that the use of combined booklets may encourage students to attempt an answer rather than write nothing. However, candidates may be wasting time if they write more in order to fill the space but this extra response is not worth extra credit. Questions from a geography AS-level past paper were arranged to form two subtests. Over 400 students each attempted one part of the test in a combined paper and one part in a separate booklet. Six students were interviewed after taking the test. Four examiners marked the scripts and completed coding sheets to record the length of responses and which of a number of options about the nature of the responses applied (e.g. Not enough depth, Evidence of 'gap-filling'/irrelevance). On both parts of the test, the mean total scores were found to be significantly higher with the combined paper than the separate booklet. The combined format often prompted longer answers and for most items elicited many more full-length answers. The combined format tended to encourage students to show their depth of knowledge and understanding, elicit fewer incidences of irrelevance (or gap-filling), reduce the frequency of good answers that were not concise and increase the occurrence of full, concise responses. The combined format gave students better information on what was needed and encouraged them to show their knowledge, understanding and skills more fully. This confirms the view that this format is preferable for exam papers requiring short to medium-length responses. (Contains 2 endnotes and 7 figures.)