This paper develops estimates of U.S. railroad productivity by using methods based on the neoclassical theory of production. We find that railroad productivity grew at the average annual rate of 1.5 percent per year during the 1951-1974 period. Using conventional measurement procedures for comparison, we find productivity growth of 3.6 percent per year. The lower estimate of 1.5 percent is the result of using procedures which better represent the railroad production process. These include using (1) estimated cost elasticities, rather than revenue shares, as output weights, (2) actual cost shares, rather than national income shares, as input weights, and (3) input and output weights which change annually.