Techniques for the estimation of skeletal muscle fiber conduction velocity are of considerable interest. These techniques use, in general, some form of cross correlation or zero-crossing analysis. Cross correlation is a straightforward method of conduction velocity estimation, however, it is difficult to realize low-cost real-time processors. Polarity correlation offers a method which preserves the advantages of cross correlation and satisfies the speed-cost constraint. This paper describes the algorithm for implementation of a polarity cofrelator instrument using a microprocessor. The instrument is tested with deterministic and stochastic signals and used to estimate the conduction velocity of biceps brachii fibers of ten normal subjects. The performance of the instrument is compared to the zero-crossing technique described by Lynn.