This paper surveys J. D. Sargan's work on instrumental variable estimation and its connections with the generalized method of moments. I first present the modelling context in which Sargan motivated instrumental variable estimation. Then I review the theory of instrumental variable estimation as developed in Sargan (1958). I discuss his approach to efficiency, his minimax estimator, tests of over- and under-identification, and his later work on the finite sample properties of IV estimators. Next, Sargan's approach to modelling IV equations with serial correlation is discussed and compared to the GMM approach. Finally I describe Sargan's (1959) results for non-linear in parameters IV models.