Abstract: Mankiw, Romer and Weil (1992) empirically established the importance of human capital accumulation in explaining the cross-country income differences. However, their results have been critiqued in the literature due to endogeneity concerns stemming from potential omitted variable bias. We re-estimate the Mankiw, Romer and Weil (1992) through generalized method of moments by using the future-time reference of languages as an instrument for physical capital, and the Protestant missionary activities in 1923 and the national primary school enrolment in 1900, as instruments for human capital. We develop an algorithm for the restricted GMM to estimate our model. We find that the original Mankiw, Romer and Weil (1992) results are robust for the method of estimation, and survive the endogeneity criticisms even when alternate indicators of human capital are considered.