In studies of human eye-head coordinated saccadic gaze shifts, different laboratories have found greatly different magnitudes of head movements for a given amplitude of gaze shift (head movement gain). The present study was conducted to examine why, and to quantify volitional head movements. Fixation/target lights were located at 20 degrees and 40 degrees on either side of a central light. There were two tasks or modes. In the non-aligned mode, gaze fixation (first light) was followed by a gaze step to the target (second light) accurately and quickly. In the head-aligned mode, the head was aligned within 3 degrees of the first fixation light (i.e., initial starting position) before the step sequence began. In both non-aligned and head-aligned modes, subject instructions pertaining to the second target light concerned only gaze; there was no requisite head position. The head movement propensity of nine subjects was ranked according to the mean gain (head/target amplitude) of two 40 degree jumps (0-40 degrees and -20 to +20 degrees) in the non-aligned mode. This ranking method clearly identified extreme head-movers and non-movers. The moderate movers were further characterized by three additional criteria, derived by comparing the gains in different jumps, which varied in starting position and amplitude. First, when the two 40 degree jumps were compared, typically the gain of non-movers was less in the symmetric jump (-20 to +20 degrees) with the gain of the head-movers was greater in the symmetric jump. Second, in the head-aligned mode the gain of non-movers progressively increased when the starting position was progressively moved eccentrically, whereas the gain of head-movers increased only slightly, if at all. Third, when the gains of two symmetric (40 degrees and 80 degrees) jumps were compared, the head-movers consistently had opposite trends from non-movers. These three comparative criteria and the initial criterion together define head movement propensity. To explain the above observations, three effects are proposed. First, a "midline-attraction" effect causes resistance to movement away from the midline in non-movers and an increase in movement amplitude if the jump starts eccentrically. Second, a "resetting" effect occurs when the eccentricity of the jump is varied; the stopping position is reset closer to the target. Third, an "awareness/arousal" effect increases the gain in the head-aligned mode due to the intrinsic nature of the alignment procedure.(ABSTRACT TRUNCATED AT 400 WORDS)