We assessed the effect of self-administration of a disease-specific health-related quality of life instrument, the Inflammatory Bowel Disease Questionnaire (IBDQ), on score results. Patients were assessed at two visits in two tertiary centers. "Experienced" patients (N = 31) with Crohn's disease had previously completed the IBDQ several times while "novices" (N = 37) with Crohn's disease or ulcerative colitis had no prior exposure to the IBDQ. At each visit a self-administered IBDQ followed by a nurse-administered IBDQ (score range, 1-7; absolute score range, 32-224) and disease activity were assessed. At visit 1, the mean rates of discrepant responses between nurse and self-administered scores were 24 +/- 15% in experienced patients and 34 +/- 17% in novice patients (p = 0.018), which fell to 21 +/- 16 and 23 +/- 10%, respectively, by visit 2 (p = NS). However, discrepancy rates were not significantly different between novice and experienced patients when adjusted by center. Discrepancies occurred randomly in all 32 IBDQ items. Eighty percent of all discrepant responses differed by only one grade of a seven-point Likert scale. Baseline self-administered scores for all patients were 4.80 +/- 1.24 (absolute score, 153.0 +/- 39.9). Mean score differences at each visit (nurse minus self) were very small, ranging from 0.029 to 0.136, and would not be considered clinically important. Intraclass correlation coefficients between the nurse and self-administered IBDQ and the four dimensional scores were > or = 0.97 by visit 2, indicating excellent concordance and minimal observer error. Mean changes in score over time were of comparable magnitude for both self (0.320 +/- 0.819) and nurse (0.260 +/- 0.831) assessments. We conclude that the IBDQ may be reliably used as a self-administered instrument in clinical trials.