The paper uses data from the World Bank's Pakistan Integrated Household Survey to study the effect of farmer schooling on the application of high-yielding variety technology in Pakistan in 1990-91. Unlike previous studies of the role of schooling in agriculture in less-developed countries, it emulates estimation of Mincer earnings functions to treat schooling as an endogenous variable. This purges the estimates of inconsistency arising from positive correlation between unobserved farmer ability and schooling, and measurement errors in observed schooling. A measure of access to schools identifies the true causal effect of schooling on the use of improved seeds. Copyright 2002 by Blackwell Publishing Ltd