The effect of introducing user fees on the frequency and quality of MEDLINE searching with GRATEFUL MED by physicians in clinical settings was tested. After training and free use (prior study), consenting participants were randomly allocated to pay searching costs (pay group) or continue without fees (no pay group). Fifty-nine physicians participated. Among the prior study's frequent searchers, the pay group searched at less than one third of the rate of those assigned to no pay. For less frequent searchers in the prior study, only 48% of those assigned to pay did any searches, compared with 85% for the no pay group (P = 0.006), and for those who did search, their frequency was almost half. However, there was no significant difference in the quality of searches; both groups demonstrated about equivalent recall (P = 0.77), but significantly lower precision (P = 0.03) than for the librarian's independent searches. Similarly, there was no difference in the proportion of searches affecting clinical decisions for the two groups. Thus, imposing user charges for online searching in clinical settings after a period of free use adversely affects searching quantity, but not quality. MEDLINE providers should consider whether user fees will undermine its benefits.