This study examined (1) the distribution of ethnic minorities among children referred to autism institutions and (2) referral bias in pediatric assessment of autism in ethnic minorities. It showed that compared to the known community prevalence, ethnic minorities were under-represented among 712 children referred to autism institutions. In addition, pediatricians (n = 81) more often referred to autism when judging clinical vignettes of European majority cases (Dutch) than vignettes including non-European minority cases (Moroccan or Turkish). However, when asked explicitly for ratings of the probability of autism, the effect of ethnic background on autism diagnosis disappeared. We conclude that the use of structured ratings may decrease the likelihood of ethnic bias in diagnostic decisions of autism. © 2008 The Author(s).