We estimate relative achievement effects of the four most commonly adopted elementary mathematics textbooks in the fall of 2008 and fall of 2009 in California. Our findings indicate that one book, Houghton Mifflin's "California Math," is more effective than the other three, raising student achievement by 0.05 to 0.08 student-level standard deviations of the Grade 3 state standardized math test. We also estimate positive effects of "California Math" relative to the other textbooks in higher elementary grades. The differential effect of "California Math" is educationally meaningful, particularly given that it is a schoolwide effect and can be had at what is effectively zero marginal cost.