The Austin Maze is a version of a popular neuropsychological test which is employed, internationally, as a measure of spatial memory. However, in Australia, failure to attain errorless performance within a limited number of trials on the Maze is often interpreted as evidence of frontal lobe dysfunction. Such a result might then be used to make predictions about a client's potential for rehabilitation or a successful return to work. Moreover, a close reading of the way in which test signs have been used to infer frontal lobe dysfunction shows that test interpretation has changed over the years. The various “frontal” formulations are reviewed in an attempt to resolve some of the ambiguity in interpretation of the Maze. One of the current “frontal” hypotheses, namelyerror utilisation, is tested with archival data from six separate samples including normal controls and heterogeneous neurological patients. These data all support the parsimonious assumption that performance on the Maze is highly predictable from errors committed over initial learning trials. Finally, some directions for future research on the assessment of frontal patients are briefly con