The stability of the neural recording is a crucial issue for cortically controlled neuroprosthetic devices. Substantial effort is devoted to developing methods to improve the chronic neural recordings. Regardless of the approach one takes to improving recording stability, the core issue is to measure just how stable recording conditions really are. In the present study, microwire array electrodes were implanted into the motor cortex area of the rhesus monkey and neural signals were recorded for more than 3 years. The data acquired on each channel were then analyzed according to spike waveform, time course of task-related activity, and preferred direction. For the time span from the beginning to the end of the recording session, the monkey was trained to do three different tasks. Initial results indicate that a certain portion of channels produces very stable recordings for over three years.