Math achievement in elementary school is mediated by performance and growth in number sense during kindergarten. The aim of the present study was to test the effectiveness of a targeted small-group number sense intervention for high-risk kindergartners from low-income communities. Children were randomly assigned to 1 of 3 groups (n = 44 in each group): a number sense intervention group, a language intervention group, or a business-as-usual control group. Accounting for initial skill level in mathematical knowledge, children who received the number sense intervention performed better than controls at immediate posttest, with meaningful effects on measures of number competencies and general math achievement. Many of the effects held 8 weeks after the intervention was completed, suggesting that children internalized what they had learned. There were no differences between the language and control groups on any math-related measures. (Contains 6 tables and 2 footnotes.)