Mutations in the juxtamembrane and kinase domains of FLT3 are common in AML, but it is not known whether alterations outside these regions contribute to leukemogenesis. We used a high-throughput platform to interrogate the entire FLT3 coding sequence in AML patients without known FLT3 mutations and experimentally tested the consequences of each candidate leukemogenic allele. This approach identified gain-of-function mutations that activated downstream signaling and conferred sensitivity to FLT3 inhibition and alleles that were not associated with kinase activation, including mutations in the catalytic domain. These findings support the concept that acquired mutations in cancer may not contribute to malignant transformation and underscore the importance of functional studies to distinguish "driver" mutations underlying tumorigenesis from biologically neutral "passenger" alterations.