This paper investigates—in the context of a simple example—the accuracy of an econometric technique recently proposed by Kydland and Prescott. We consider a hypothetical econometrician who has a large sample of data, which is known to be generated as a solution to an infinite horizon, stochastic optimization problem. The form of the optimization problem is known to the econometrician. However, the values of some of the parameters need to be estimated. The optimization problem—presented in a recent paper by Long and Plosser—is not linear quadratic. Nevertheless, its closed form solution is known, although not to the hypothetical econometrician of this paper. The econometrician uses Kydland and Prescott’s method to estimate the unknown structural parameters. Kydland and Prescott’s approach involves replacing the given stochastic optimization problem by another which approximates it. The approximate problem is a element of the class of linear quadratic problems, whose solution is well-known—even to the hypothetical econometrician of this paper. After examining the probability limits of the econometrician’s estimators under “reasonable” specifications of model parameters, we conclude that the Kydland and Prescott method works well in the example considered. It is left to future research to determine the extent to which the results obtained for the example in this paper applies to a broader class of models.