A knowledge of basic statistical concepts is becoming increasingly important for a proper understanding of much of the medical literature. Unfortunately, most medical schools include an introductory biostatistics course only in the preclinical years. During clinical rotation and residency training, when a physician would most appreciate the need for statistical concepts, it is rarely a part of the curriculum. Moreover, even what was learned during the earlier part of medical training tends to be forgotten by the time these concepts are needed most. To organize future educational seminars and to evaluate the need for education in biostatistics, we conducted a survey among a group of pediatricians attending a review course held in August 1987. A questionnaire containing a set of 16 problems on basic medical statistical concepts, similar to those asked by Wulff et al? in a 1983 survey of Danish physicians, was distributed. More than 50% of the 68 respondents answered 13 of the 16 questions incorrectly, and only three chose correct answers to 13 or more questions. The median number of correct answers was seven. The median number of correct answers from the 34 physicians who graduated from medical schools within the past 9 years was 8.5, compared with 5.0 for the 30 who graduated 10 or more years ago (a median test for independence yielded a value of P <0.042). Incorrect answers were most frequent to questions on regression tests, analysis of variance, and the difference between standard deviation and standard error. The results of this survey reaffirm a need for continuing