A recently proposed Nordic Sleep Questionnaire (NSQ) comprises 26 questions concerning qualitative and quantitative aspects of the respondent's sleep habits. Its reproducibility was evaluated in 32 spinal cord injured individuals (SCI), 24 men and eight women (23-72 years), and 79 normal subjects, 23 men and 56 women (19-77 years). They completed the NSQ twice at a median interval of 15 days (range 10-26) and 27 days (range 4-103) respectively. The group of normal subjects were evenly divided into group 26, i.e. those who completed the two NSQs within 26 days, and group 27 with 27 days or more between their replies. Generally, group 27 showed no worse test-retest agreement than group 26. In addition, the respondents' answers, with a few exceptions, were reasonably stable in terms of test-retest agreement or standard deviation. The SCI group exhibited the same level of reproducibility, although they had more 'pathology' to report and thus more scope for contradicting themselves. The questions in the NSQ generally were satisfactorily reproducible. However, answers to the ordered five-point questions about sleepiness in the morning and during the daytime ought to be interpreted with caution. The same may be said about the number of minutes required to fall asleep, and the duration of daytime naps.