We examine the degree to which very experienced and inexperienced auditors can detect the presence versus absence of deception from the Q&A portions of earnings conference calls. As predicted, experienced auditors outperform chance and inexperienced auditors. This is encouraging, as recent meta-analyses from social psychology document that experts generally outperform neither chance nor novices in detecting deception. More worrisome, but also as predicted, experienced auditors’ performance edge predominantly stems from fewer false positives. While false positives decrease audit efficiency, false negatives jeopardize audit effectiveness, which is what financial statement users likely care more about. We also manipulate whether auditors have only transcript excerpts or these plus related audio. Audio improves accuracy more for inexperienced than experienced auditors, as we predict. In supplementary tests, we show that combining judgments of several inexperienced auditors who received audio improves judgment accuracy to the point of being on par with the average accuracy of experienced auditors. Thus, labor substitution could help audit firms get relatively accurate deception predictions with lower costs. Finally, we investigate process measures underlying the auditors’ fraud judgments. We find experienced auditors do identify more red flags, but that their propensity to locate red flags is positively correlated only with their fraud judgments and not with the actual incidence of fraud.