This study examined the ability of four measures of suboptimal performance to correctly classify four groups of subjects (normal controls, uncoached malingering, coached malingering, and head injured). Only the Portland Digit Recognition Test-Computerized (PDRT-C) identified simulating malingerers with greater than chance accuracy while minimizing false positives. Coached subjects were better able than their uncoached counterparts to avoid detection on all measures. In an additional analysis, a discriminant function using the response latency and total correct scores from the PDRT-C identified 70% of the coached malingerers on cross validation. The three other tests (Nonverbal Forced Choice Test, 21-Item Test, and Dot Counting Test) failed to obtain a satisfactory classification rate for the malingering groups as a whole and coached malingerers in particular.