Medical librarians and informatics professionals believe the medical journal literature can be useful in clinical practice, but evidence suggests that practicing physicians do not share this belief. The authors designed a study to determine whether a random sample of "native" questions asked by primary care practitioners could be answered using the journal literature. Participants included forty-nine active, nonacademic primary care physicians providing ambulatory care in rural and nonrural Oregon, and seven medical librarians. The study was conducted in three stages: (1) office interviews with physicians to record clinical questions; (2) online searches to locate answers to selected questions; and (3) clinician feedback regarding the relevance and usefulness of the information retrieved. Of 295 questions recorded during forty-nine interviews, 60 questions were selected at random for searches. The average total time spent searching for and selecting articles for each question was forty-three minutes. The average cost per question searched was $27.37. Clinician feedback was received for 48 of 56 questions (four physicians could not be located, so their questions were not used in tabulating the results). For 28 questions (56%), clinicians judged the material relevant; for 22 questions (46%) the information provided a "clear answer" to their question. They expected the information would have had an impact on their patient in nineteen (40%) cases, and an impact on themselves or their practice in twenty-four (51%) cases. If the results can be generalized, and if the time and cost of performing searches can be reduced, increased use of the journal literature could significantly improve the extent to which primary care physicians' information needs are met.