A seven sensor array (6 sensors equally spaced on a 4 cm dia. with 1 sensor at the center) was used to differentiate two close sensorimotor sources (voluntary movements of right index finger and right thumb). Magnetic field amplitude data was analyzed from a single-session recording (single sensor-array location). The sensor array was positioned to record the peak magnetic field amplitude (an extremum) of one or both digits (analyzed at 145 ms after the electromyogram). An extremum was obtained by plotting a topographic field map and also by mathematical interpolation (both) from the seven-sensor amplitudes while at a fixed location. This method of analysis graphically showed digit extrema separation in all subjects over the negative extrema and thus separation of the neural generator populations at this specific time. Statistical significance of the seven sensor amplitudes was examined by null hypotheses testing. The digit extrema separation (over the negative extrema) was influenced by errors primarily due to noise, signal-to-noise ratio = 6 +/- 2.4, and the subjects' variability in digit flexion. This analysis eliminated sensor array position error and minimized the analysis time to a single session. From the plotted distribution of extrema pertaining to a particular flexion (same subject), obtained from repeated sessions, a mean extrema location for each digit was determined. In this manner, a separation between the digit extrema was demonstrated. The mapped distribution also provided a clear display of the error in plotting extrema locations. For both digits the error in mapping extrema locations (negative only) was 0.9 cm.(ABSTRACT TRUNCATED AT 250 WORDS)