Using a basic mathematics competence test based on the primary school curricular standard, we examine the extent to which years spent in school actually increases numeracy achievement in rural Bangladesh. Our sample includes 10-18-year-old children currently enrolled in school as well as those out of school. About half of the children failed to pass the written competence test, a finding that also holds for those having completed primary school. Even after holding constant a wide range of factors such as household income, parental characteristics, current enrollment status, child ability, and aggregate institutional indicators of school quality, there remains a statistically significant correlation between schooling attained and basic mathematics competence above and beyond primary school completion--but the estimated schooling-learning profile is relatively flat. The findings have wide implications for reformulating policies that tend to focus on quantitative expansion of education in developing countries, without concurrent improvements in learning.