There is a perception amongst health care professional that patients under-report their smoking habits. The aim of this study was to validate self-reported smoking habits in patients who have been treated for an oral cancer using saliva cotinine. In a cross-sectional study100 consecutive patients attending a maxillofacial oncology clinic completed a smoking related questionnaire following which a saliva sample was obtained. Saliva cotinine levels were determined by gas-liquid chromatography. The mean (SD) age 61 (11), 74% male, 26% female. The majority (79%) had Stage I/II disease, which were treated by surgery (49%), radiotherapy (14%) or combined therapy (37%). Average time (SD) since diagnosis was 28 (24) months. 42% were self-reported smokers. Cotinine assessment was possible from 91 patients. Of these 43% (39/91) were smokers by self-report, all were biochemically smokers (cotinine level>14 microg/l). 9.6% (5/52) patients who claimed to be non-smokers by self-report had cotinine levels suggesting recent active smoking. The level of agreement was excellent (kappa = 0.89), and specificity and sensitivity high (1 and 0.90, respectively). Self-reported smoking habits are reasonably accurate in this group of patients. We believe that smoking related research using self-report alone can reliably be carried out in this particular patient group.