1) This study investigates the early development of adaptive changes in oculomotor function associated with coordinated eye-head tracking of the optically reversed image of an earth-fixed target seen through horizontally reversing dove prism goggles attached to the skull. 2) Two tasks comprised a) fixation of a single target during head rotation which causes the seen target's image to move in the direction of head motion by an amount exactly equal to the head movement itself (the 1-Target task), and b) change of gaze onto a displaced target with head free to move (2-Target task). 3) The 1-Target task requires the eyes to move in a direction opposite to that of the normal vestibulo-ocular reflex (VOR). The 2-Target task is identical, except that reorientation onto the new target calls for an initial saccadic eye movement in a direction opposite to that of the ensuing head movement, which is contrary to the normal pattern of eye-head coordination during gaze shifts. 4) Eye (EOG) and head (potentiometer) movements were continuously recorded (0-250 Hz) in an apparatus which permitted sudden, unexpected, electromagnetic braking of the head movement, either just before or during the intended manoeuvre. 5) Early adaptive strategies employed reduction of VOR gain, rearrangement of timing, amplitude and shape of "catch-up" saccades and the introduction of centrally programmed eye movements uncovered by the braking manoeuvres. 6) All of these phenomena were detectable in an initial series of 60 trials, in which the total exposure to visual-vestibular conflict was less than 30 s. They became more systematized and more marked after 6 h of active reversed vision experience. 7) Specifically, mean VOR gain, measured within the first 80 ms of head movement (deemed free of visuomotor influence), became markedly attenuated (25% in the first test series; 66% after 6 h of active vision-reversed exercise). In addition (not included in the above percentages) there were numerous occasions of complete absence of measurable VOR during head rotation, in both the first and final test series. 8) In the 1-Target task, the latency of the first "catch-up" saccade (re onset of head movement) tended to offset residual VOR by becoming shortened to the point of synchrony with head movement onset. This saccade (not present in control tests) continued to occur on those occasions when the head was unpredictably prevented from moving, and when head movements were made in the dark.(ABSTRACT TRUNCATED AT 400 WORDS)