This study examined the changing role and longitudinal predictive validity of curriculum-embedded progress-monitoring measures (CEMs ) for kindergarten students receiving Tier 2 intervention and identified as at risk of developing reading difficulties. Multiple measures were examined to determine whether they could predict comprehensive latent first- and second-grade reading outcomes and whether their predictive validity changed concurrent with reading development. CEMs of phonemic, alphabetic, and integrated tasks were given 3 times during the kindergarten year to 299 students. Structural equation modeling indicates that CEMs explained a significant amount of variance on first- (54%–63%) and second-grade (34%–41%) outcomes. The predictive validity of specific measures varied over the kindergarten year with sound and letter identification measures being predictive early and segmenting and word reading becoming important as reading abilities progressed. Findings suggest that CEMs may be viable and helpful tools for making data-driven instructional decisions in a response to intervention framework.