One clinical use of the ophthalmometer is to aid in the prediction of the correcting cylinder for the eye, thereby to provide a starting point for retinoscopic and/or subjective testing. When axis is known in advance with reasonable accuracy, correcting cylinder refinement is accomplished easily, but when the predicted axis is in substantial error, perhaps by as much as 90°, then a starting point has little value and refinement gives way to primary determination. For optical reasons, the accuracy of predicting the correcting cylinder axis increases with corneal toricity. Therefore, the clinical value of the ophthalmometer as a predictor of the correcting cylinder increases with escalating corneal toricity. The procedure followed in this paper has been first to assess, through Monte Carlo simulation, the accuracy of the prediction of correcting cylinder axis as a function of direction and amount of corneal astigmatism. I then considered the effects of potential errors in analysis that might result from uncertainty as to the actual values of certain parameters. Greatest uncertainty appears to attend variance of the axis of the residual astigmatism. It is suggested that the minimum amounts of astigmatism which provide reasonable assurance that the axis of the correcting cylinder will lie within 10 of its ophthalmometrically predicted value at least 90% of the time are: (1) 2.25 D, for W-T-R astigmatism; (2) 1.50 D, for A-T-R astigmatism; and (3) 2.25 D, for oblique astigmatism.