There are conflicting reports on the effect of vergence training. In two studies using push up and prism vergence exercises the conflict is shown to result from differences in the constancy of effort. Improvement only occurs in exercises involving sustained effort in the direction being trained. The effect of alternating movements in both directions is small and inconsistent. Substantial longlasting gains in either convergence or divergence prism vergence scores can result from just 5 min of sustained effort at an angle halfway between the break and recovery points. Results for adults and children on motor-driven instruments are comparable and the scores obtained discriminate better than those from hand-turned instruments. The most likely physiological model involves potentiating processes which are specifically maximized by isometric exercise. Potentiation results in a positive feedback process, the effects of which are strong enough to precipitate strabismus and which might be deliberately manipulated in therapy of poor vergence control, if responses in the required direction can be induced.