ObjectiveTo assess a question–answer pair (QAP) database integrated with websites developed for drug information centres to answer complex questions effectively.DesignDescriptive study with comparison of two subsequent 6-year periods (1995–2000 and 2001–2006).SettingThe Regional Medicines Information and Pharmacovigilance Centres in Norway (RELIS).ParticipantsA randomised sample of QAPs from the RELIS database.Primary outcome measureAnswer time in days compared with Mann–Whitney U test.Secondary outcome measureNumber of drugs involved (one, two, three or more), complexity (judgemental and/or patient-related or not) and literature search (none, simple or advanced) compared with χ2 tests.Results842 QAPs (312 from 1995 to 2000 and 530 from 2001 to 2006) were compared. The fraction of judgemental and patient-related questions increased (66%–75% and 54%–72%, respectively, p<0.01). Number of drugs and literature search (>50% advanced) was similar in the two periods, but the fraction of answers referring to the RELIS database increased (13%–31%, p<0.01). Median answer time was reduced from 2 days to 1 (p<0.01), although the fraction of complex questions increased from the first to the second period. Furthermore, the mean number of questions per employee per year increased from 66 to 89 from the first to the second period.ConclusionsThe authors conclude that RELIS has a potential to efficiently answer complex questions. The model is of relevance for organisation of drug information centres.