A variety of methods for estimating premorbid intelligence were compared in an Australian sample participating in the Macquarie University Neuropsychological Normative Study (MUNNS). Predictors included a contextual version of the AUSNART, the WRAT-3 Reading subtest, Spot-the-Word Test and sub-tests from the WAIS-R. The study involved 244 healthy participants aged between 18 and 34 years, with half of them serving as a cross-validation sample. Multiple regression was used for all analyses. When taking into account both the statistical results and clinical utility of the methods, the Sydney Premorbid Intelligence Estimate (SPIE), consisting of Vocabulary and Picture Completion subtests from the WAIS-R, was considered to be the best predictor. Combining all measures into one regression did not substantially increase the amount of variance explained. Base rate tables for predicted IQ versus obtained IQ, and the discrepancy between predicted 10 and Delayed Recall Index of the WMS-R are provided for each of the prediction methods. Further research to examine the clinical utility of these methods is required.