At our university, Internal Medicine clerks are members of a team responsible for the care of patients hospitalized on a teaching ward. Clerks first encounter their patients after the latter have been fully worked up by other physicians who have examined them and initiated investigations and management. Clerks are thus deprived of the opportunity to practice information acquisition, hypothesis generation and problem solving. We therefore undertook a 'blinding' initiative wherein each clerk was required to work up at least one hospitalized patient per week without access to the patient chart and without knowledge of information acquired and hypotheses generated by other physicians. Weekly data collection during the 8-week experiment with 40 clinical clerks revealed that work up of 'blinded' patients was more time-consuming and more difficult than work up of unblinded patients. Clerks were appreciative of the educational value of blinding. Teaching faculty felt clerk 'blinding' to be a practical approach to approximating the true conduct of medical practice and as such was useful for student learning.