Background: Children's vocabulary knowledge is closely related to other measures of language development and to literacy skills and educational attainment. Aim: To use a regression discontinuity design (RDD) to evaluate the effectiveness of a small-group vocabulary intervention programme for children with poor vocabulary knowledge. Methods & Procedures: The vocabulary knowledge of children (N = 199) aged 6-9 years was assessed in six classes. Based on scores at the initial assessment, children with low vocabulary scores for their age were assigned to an intervention group (43 children), with the remaining 156 children assigned to a control group. Children in the intervention group received two to three small-group weekly teaching sessions over a 10-week period. All children were retested post-intervention on the same measures of vocabulary knowledge. Outcomes & Results: The intervention group showed significant improvements in their knowledge of the meanings of the taught words at post-test (an additional 3.95 words learned [95% confidence interval (CI) = 2.70-5.20] compared with the control group; d = 1.20), but the effects did not generalize to untaught words. Conclusions & Implications: A small-group vocabulary intervention programme is effective for teaching word meanings to 6-9-year-old children with poor vocabulary skills. This study provides further evidence that the RDD is an effective method for evaluating educational interventions.