A method is described for estimating the distribution of nerve-fiber conduction velocities in a nerve bundle. This method is based on a detailed general model of the nerve bundle compound action potential, which is characterized as a weighted sum of delayed single-fiber action potentials. The non-iterative estimation method is applied to two examples taken from existing literature, demonstrating the similarity of conduction velocity and fiber diameter distributions, sensitivity of the estimate to variations in important model parameters, and applicability to the differentiation of normal and abnormal nerve function.