OBJECTIVE: Noninvasive brain stimulation in primary progressive aphasia (PPA) is a promising approach. Yet, applied to single cases or insufficiently controlled small-cohort studies, it has not clarified its therapeutic value. We here address the effectiveness of transcranial direct current stimulation (tDCS) on the semantic PPA variant (sv-PPA), applying a rigorous study design to a large, homogeneous sv-PPA cohort. METHODS: Using a double-blind, sham-controlled counterbalanced cross-over design, we applied three tDCS conditions targeting the temporal poles of 12 sv-PPA patients. Efficiency was assessed by a semantic matching task orthogonally manipulating "living"/"nonliving" categories and verbal/visual modalities. Conforming to predominantly left-lateralized damage in sv-PPA and accounts of interhemispheric inhibition, we applied left hemisphere anodal-excitatory and right hemisphere cathodal-inhibitory tDCS, compared to sham stimulation. RESULTS: Prestimulation data, compared to 15 healthy controls, showed that patients had semantic disorders predominating with living categories in the verbal modality. Stimulation selectively impacted these most impaired domains: Left-excitatory and right-inhibitory tDCS improved semantic accuracy in verbal modality, and right-inhibitory tDCS improved processing speed with living categories and accuracy and processing speed in the combined verbal × living condition. INTERPRETATION: Our findings demonstrate the efficiency of tDCS in sv-PPA by generating highly specific intrasemantic effects. They provide "proof of concept" for future applications of tDCS in therapeutic multiday regimes, potentially driving sustained improvement of semantic processing. Our data also support the hotly debated existence of a left temporal-pole network for verbal semantics selectively modulated through both left-excitatory and right-inhibitory brain stimulation. Ann Neurol 2016;80:693-707.