PurposeSeveral tau PET tracers have been developed, but it remains unclear whether they bind to the same molecular target on the heterogeneous tau pathology. In this study we evaluated the binding of two chemically different tau-specific PET tracers (11C-THK5351 and 11C-PBB3) in a head-to-head, in vivo, multimodal design.MethodsNine patients with a diagnosis of mild cognitive impairment or probable Alzheimer’s disease and cerebrospinal fluid biomarker evidence supportive of the presence of Alzheimer’s disease brain pathology were recruited after thorough clinical assessment. All patients underwent imaging with the tau-specific PET tracers 11C-THK5351 and 11C-PBB3 on the same day, as well as imaging with the amyloid-beta-specific tracer 11C-AZD2184, a T1-MRI sequence, and neuropsychological assessment.ResultsThe load and regional distribution of binding differed between 11C-THK5351 and 11C-PBB3 with no statistically significant regional correlations observed between the tracers. The binding pattern of 11C-PBB3, but not that of 11C-THK5351, in the temporal lobe resembled that of 11C-AZD2184, with strong correlations detected between 11C-PBB3 and 11C-AZD2184 in the temporal and occipital lobes. Global cognition correlated more closely with 11C-THK5351 than with 11C-PBB3 binding. Similarly, cerebrospinal fluid tau measures and entorhinal cortex thickness were more closely correlated with 11C-THK5351 than with 11C-PBB3 binding.ConclusionThis research suggests different molecular targets for these tracers; while 11C-PBB3 appeared to preferentially bind to tau deposits with a close spatial relationship to amyloid-beta, the binding pattern of 11C-THK5351 fitted the expected distribution of tau pathology in Alzheimer’s disease better and was more closely related to downstream disease markers.Electronic supplementary materialThe online version of this article (10.1007/s00259-018-4012-5) contains supplementary material, which is available to authorized users.