We examined whether scientific reviewers exhibit bias in scoring a simulated "positive" study (i.e. showing adverse fetal effects) as compared to a simulated "negative" study on the fetal effects of binge drinking. The reviewers of the "negative" study tended to reject it more commonly, to give it lower scores, and there was significantly more variability from the median in their scores. Scientific journals should make an effort to eliminate this source of bias against negative results.