Elementary math instruction traditionally has emphasizedprocedures rather than concepts. Thus, students tend to lack astrong understanding of foundational concepts likemathematical equivalence. Cognitive scientists andmathematics educators have found small yet effective ways tomodify traditional arithmetic instruction to promote students’conceptual understanding of math equivalence. Educationalstandards also now reflect this academic research. However, itis unclear whether classroom practices have caught up withresearch and policy. In the current study, we observed teachers’practices during arithmetic instruction. The goal was todetermine if teachers are using research-based practices thatpromote understanding of math equivalence and if variation inuse of research-based practices is associated with students’growth in understanding of math equivalence across the schoolyear. Eight second and third grade classrooms (M students perclassroom = 23) were observed twice during math instruction.Students completed a math test both before and after theobservation period. Research-based practices were rarelyobserved in any classrooms, so there was not much variation inclassroom use of research-based practices to predict studentgrowth. Students improved their performance on all problemtypes tested, but performance on math equivalence problemswas significantly lower than on other problem types. Resultssuggest that policies and practices designed to improvestudents’ understanding of math equivalence may not havefiltered down to affect instructional practices in classrooms.