The construction and validation of a tutor evaluation questionnaire are described. The importance of the tutor in problem-based learning requires assessment of staff conducting this role. After a pilot study in 1990-91 with 100 students and 150 tutors who identified important tutor skills, a questionnaire was prepared. This questionnaire was tested with six courses in the 1991-92 school year, and improvements were made based on the results. The pilot studies resulted in a list of 13 Likert scale statements used in an instrument administered to about 1,079 students in 8 courses (142 tutorial groups) at the Maastricht Medical School (Netherlands) in the 1992-93 academic year. Students indicated whether their tutors demonstrated the behavior described in each statement. Confirmatory factor analysis shows that a 3-factor model comprising 13 items does not fit the data, although skipping 2 items improved the model's fit. Nevertheless, these two items were considered important for content validity. Alpha coefficients indicate that questionnaire reliability is high. Findings otherwise indicate that the students' global and analytical ratings about the functioning of the tutor correspond highly with each other. Some problems involved in administering the questionnaire are discussed. Two tables present study findings. (Contains 11 references.) (SLD)