Background: Technology advances make it increasingly possible to adapt direct behavioral assessments for classroom use. This study examined children's scores on HTKS-Kids, a new, largely child-led version of the established individual research assessment of self-regulation, Head-Toes-Knees-Shoulders-Revised task (HTKS-R). For the HTKS-Kids tablet-based assessment, which was facilitated by children's preschool teachers, we examined (1) preliminary reliability and validity; (2) variation in scores predicted by child age and background characteristics; and (3) indication that HTKS-Kids provides different information from teacher ratings of children. Method: Participants included n = 79 4-year-old children from two urban areas in upstate New York, USA. Average parent education was 12.5 years, ranging 3–20. A researcher administered the HTKS-R to individual children, and teachers (eight white, two Latino) were trained to use the HTKS-Kids tablet-based assessment and asked to play once with each study child. Teachers also rated each child on 10 Child Behavior Rating Scale (CBRS) items about classroom self-regulation. Results: We found evidence that (1) the HTKS-Kids captures variation in children's self-regulation and correlates positively with established measures, (2) parent education was the best predictor of HTKS-Kids scores, and (3) teachers rated Black children significantly worse and white children better on the CBRS, with the magnitude of group differences similar to the contribution of parent education. In contrast, Black and white children showed no score differences on HTKS-Kids. Implications: The HTKS-Kids is a promising new tablet-based assessment of self-regulation that could replace or supplement traditional teacher ratings, which are often subject to implicit bias. (author abstract)