In this systematic review and meta-analysis of 19 studies involving 2568 cytologically indeterminate thyroid nodules, Valderrabano et al assess whether the diagnostic performance of a common thyroid nodule gene expression classifier in the initial validation study was consistent with results of postmarketing studies. They found that the initial validation study cohort was not representative of the populations in which the gene expression classifier had been used, calling into question its reported diagnostic performance, including its negative predictive value. Marti and Shaha provide an Invited Commentary.