The assessment of soft contact lens fit by observation of the lens behavior on the eye is one of the most common procedures undertaken in contact lens practice. Nevertheless, the guidelines for doing so have been derived empirically with little or no research into the usefulness of these observations. Over 2000 evaluations of soft contact lens fit were analyzed to determine the predictive value of various indicators of contact lens fit, namely: subjective comfort, lens centration, postblink movement (PBM), version and upgaze lag, and tightness assessed by the push-up test. The receiver-operator characteristic (ROC) curves were prepared for each variable by plotting sensitivity values against false positive rates (FPRs) for various cut-off points. Sensitivity was defined as the proportion of “bad” fits correctly identified by a given variable; FPR was defined as the proportion of “good” fits incorrectly identified as bad fits. With each fitting variable, the optimum cut-off point and its corresponding sensitivity and FPR were determined from the ROC curves. Subjective comfort was found to be of little value in identifying tight fitting lenses but was of some limited value in identifying loose fitting lenses, particularly high water content (HWC) lenses. The push-up test was the most accurate single test of lens fit acceptability having a high sensitivity for both tight (92%) and loose (80%) lens fits. Assessment of PBM was a sensitive indicator of tight fitting lenses (95%), but was a poor indicator of loose fitting lenses, having a low sensitivity (51%). Assessment of version lag showed better sensitivity than upgaze lag in assessing loose fits, whereas upgaze lag showed better sensitivity for tight lens fits. Lens centration was of no predictive value in identifying tight fitting lenses but, as expected, was of some value in identifying loose fitting lenses. The results of this study enable a critical reassessment of the procedures that practitioners follow when fitting soft contact lenses.