In this meta-analysis of 54 longitudinal studies with over 58,000 students in kindergarten through 12th grade, we examined the predictive nature of early numeracy measured at or before the first year of formal schooling in relation to later mathematics. Results showed that early numeracy significantly predicted mathematics measured after 6 months or later, r = 0.49, 95% confidence interval [0.47, 0.52]. After controlling for all moderators in a model, results indicated that (a) different early numeracy including numbering, relations, and arithmetic operations did not differ much in their predictions of different later mathematics; (b) early numeracy as a whole was more predictive of later advanced mathematics skills (word problems) than of later foundational mathematics skills (calculations and fact fluency); (c) early numeracy's prediction of later mathematics was stronger with longer prediction intervals; and (d) the earlier early numeracy was assessed, the stronger its prediction of later mathematics. Together, these findings suggest that early numeracy may be a unitary construct. Early numeracy does not merely serve as a steppingstone with temporary effects on foundational mathematics; instead, it likely triggers a snowballing effect, cumulatively influencing mathematics development over time. [This is the online first version of an article published in "Journal of Educational Psychology" (ISSN 0022-0663).]