This paper exploits a unique long-horizon longitudinal data set fromTanzania to examine the long-run consequences of child labor oneducation, employment choices, and marital status. Using crop shocksas instruments, our 2SLS estimates indicate that child labor is causally associated with reduced educational attainment (both as measured by the number of school years as well as by an indicator capturing completion of primary school). Interestingly, this result appears to be entirely driven by the sample of boys, for whom doubling labor hoursfrom a mean prevalence (16 hours) would imply losing 80% of a school year. Boys who worked when young are more likely to be farming (as opposed to earning a wage), although we could not find evidence that child labor is associated with noticeable differences inthe choices of crop (cash versus subsistence) or with subsequent migration. For girls, the main discernable effect is on early marriage: a higher level of child labor hours is associated with a substantially greater chance of being married 10 to 13 years later.