The Edinburgh Handedness Inventory was administered to a sample of 203 mentally well adults drawn from the Western Australian Family Study of Schizophrenia (90 men and 113 woman). Confirmatory factor analysis demonstrated that seven out of ten original items of the Edinburgh Handedness Inventory were sufficient to provide an internally consistent and valid measure of hand preference. Exclusion of three problematic items led to a more refined measurement of the latent construct of handedness. The rationale for exclusion was: (1) redundancy stemming from collinearity between writing and drawing, and (2) an unacceptably large measurement error associated with two of the items (use of broom and opening a box-lid). The results suggest that this revision of the Edinburgh Handedness Inventory enhances its measurement properties.