Correlation and regression analyses were performed on thirty-three of the magnocellular red nucleus cells described in the previous paper. We sought to test for reliable relations between the parameters of individual tracking movements and corresponding bursts of neural discharge. High correlations were found between the following burst and movement parameters: (i) burst latency versus movement latency; (ii) burst duration versus movement duration; (iii) burst frequency versus movement velocity and (iv) number of spikes in the burst versus movement amplitude. Cells were ranked according to the average of the duration, velocity and amplitude correlation coefficients. The top twenty cells had average correlation coefficients ranging from 0.69 to 0.88 for their preferred movement. These cases were judged most likely to reveal the control functions of the red nucleus, and the following points refer to this sample. Burst onset led movement onset by 118 +/- 23 ms, and burst offset led movement offset by 50 +/- 38 ms. Burst duration increased as the duration of the movement increased (r = 0.87 +/- 0.11). The duration of the burst was approximately equal to movement duration (slope of 0.99 +/- 0.16) plus a constant (72 +/- 34 ms) throughout a broad range. Average discharge rate during the burst increased with average movement velocity (r = 0.69 +/- 0.15). The slope of the relation was 0.36 +/- 0.21 (pulses/s)/(deg/s) of joint rotation. The regression lines had consistent upward offsets (56 +/- 15 pulses/s) that exceeded the spontaneous discharge rate (17 +/- 10 pulses/s). The number of spikes in the burst increased with movement amplitude independent of velocity (r = 0.72 +/- 0.11). The slope of the relation was 0.62 spikes/deg and the offset was 13 +/- 4 spikes. The preferred movement was co-ordinated hand in fifteen cases, digit in three, elbow in one and shoulder in one. When these cells were tested with an alternate movement, the failure rate (cases in which a burst did not accompany a movement) increased from 1.4 to 20%, and the correlation coefficients generally were low and lacked significance. Cells in the top twenty had directionally specific responses, low variance in lead time, large depths of modulation (41-118 pulses/s) and low failure rates. Cells that failed to show strong parametric correlations often had one or more of the former attributes. It appears that high parametric correlations with individual movements are particularly restrictive criteria of relatedness.(ABSTRACT TRUNCATED AT 400 WORDS)