Using data from the Early Childhood Longitudinal Study (ECLS) 1998/1999, we examine the relationship between elementary students' (K-5) manipulative use and mathematics learning. Using a cross-sectional correlational analysis, we found no relationship between manipulative use and student mathematics "achievement". However, using a longitudinal analysis, we documented a positive relationship between manipulative use and student mathematics "learning" during their elementary school years (K-5). From a teaching and learning perspective, these findings provide important evidence of the influence of long-term manipulative use on students' overall learning. From policy and methodological perspectives, these findings provide evidence for the importance of modeling student learning (as opposed to achievement) when studying the effectiveness of instructional strategies.