The Minnesota Multiphasic Personality Inventory (MMPI) and, more recently, its revised version (the MMPI-2) have represented the "gold standard" in the psychometric assessment of malingering and other response styles. In this study, we provide a stringent test of the MMPI-2 validity indices and their ability to detect feigned schizophrenia in four groups of simulators (n = 72). Simulators were randomly assigned to one of four conditions: (a) coached on symptoms of schizophrenia, (b) coached on strategies for the detection of fakers, (c) coached on both symptoms and strategies, or (d) uncoached. Simulators were compared to subjects responding under an honest condition (n = 13) and a comparison group of schizophrenic inpatients (n = 37). We found knowledge of strategies alone allowed many simulators (i.e., one third or more, depending on the validity indices) to elude detection. In contrast, knowledge of the disorder appeared less useful to simulators in avoiding detection. Coaching on both strategies and symptoms was not as effective as strategies alone. Consistent with previous studies, uncoached simulators were detected with moderately high levels of accuracy.