PURPOSE. To evaluate the educational effect on residents of a grand rounds given by a pharmaceutical company employee. METHOD. Using a retrospective cohort study design, the authors questioned 75 housestaff at a university hospital three months after a February 1990 grand rounds on Lyme disease to determine whether the residents' beliefs about the drug of choice for this disease differed between attendees and non-attendees. Odds ratios, 95% confidence intervals, and logistic regression were used for the analysis of results. RESULTS. The 22 housestaff who had attended the grand rounds were more likely to choose appropriately the cephalosporin manufactured by the speaker's company over other drugs for patients with Lyme disease presenting with second-degree heart block (adjusted odds ratio of 8.4; 95% CI 2.1-38.9). However, they also chose it inappropriately for first-degree heart block (adjusted odds ratio of 7.8; 95% CI 1.6-45.5). None of the attendees, compared with 11 (21%) of the non-attendees, named an oral antibiotic for both of two milder presentations, even though oral therapy would be more appropriate (p = .027). CONCLUSION. The results suggest that grand rounds effectively change residents' beliefs, but a sponsoring company's drug may be favored. Information assimilated in this way may not be well supported by the scientific literature and could result in a choice of treatment that is more expensive than other acceptable treatments.