Response biases are systematic biases in responding to test items that are unrelated to the content of the items. Examples often reported in young children include choosing only the lowest or highest anchors of a scale, or choosing a left-to-right sequence of responses. We investigated the presence of response biases in young children's ratings of pain in hypothetical situations, as a way of gauging their developing understanding of a pain scale over the preschool years. Children aged 3-5 years (N=185) rated items from the Charleston Pediatric Pain Pictures (CPPP) using the Faces Pain Scale-Revised (FPS-R). Response biases were identified objectively by computer pattern identification. Anchor biases (choosing the lowest and highest pain faces) occurred in 16% of children. Left-right or right-left sequences occurred in 35%. Monte Carlo simulation established that such patterns occur infrequently by chance (<3% for anchor biases; <6% for sequence biases). Response biases were identified more often in younger than older children. These results reveal that response biases are common in children under 5 years. Clinicians should consider self-report pain ratings from preschoolers with caution, seek complementary observational assessment, and investigate discrepancies between self-report and observational estimates of pain. Simplified forms, instructions, and methods of administration for self-report scales should be developed and validated for use with 3- and 4-year-olds.