Background: The pathological mechanisms underlying Alzheimer’s disease (AD) are still not fully understood. While amyloid (Aβ) deposition and tau aggregation are cardinal features, microglial activation is also an important component. However, the relationship between the three processes is widely debated. This study used positron emission tomography (PET) imaging to evaluate this relationship in a cohort of mild cognitive impairment (MCI) and Alzheimer’s disease (AD) subjects. Methods: Nineteen healthy controls, thirty-nine MCI and nineteen AD individuals had neuropsychological testing, MRI, 18F-flutemetamol and 11C-PBR28 PET. Eight controls, twenty-five MCI and eighteen AD individuals also had 18F-AV1451 PET. For 18F-flutemetamol and 18F-AV1451, target: cerebellar ratio images were used to calculate standardised uptake value ratio (SUVR) values. For 11C-PBR28, a two-tissue compartment model and Logan graphical analysis were used to determine volumes of distribution (VT). Group comparisons, and correlations between the tracers, were examined at regional and voxel level. Results: 1) Mean 11C-PBR28 binding was significantly higher in the Aβ-positive MCI group compared to the controls at a voxel level, while there was no difference between AD and controls. However, 42% of Aβ-positive AD individuals had significantly higher binding than the healthy control mean+ 2 standard deviations 2) All three processes were correlated at a voxel level; microglial activation and tau aggregation were correlated in connected regions on analysis 3) Aβ and tau load were correlated at a voxel and regional level; in addition medial temporal lobe tau binding correlated with isocortical Aβ load 4) Certain Aβ-negative individuals also demonstrated significantly increased levels of microglial activation and tau aggregation compared with controls on single-subject analysis. Conclusions: In this study examining inter-relationships between Aβ deposition, tau aggregation and microglial activation in MCI and AD, significant positive correlations were found between all three processes. These correlations were both local (overlapping voxels) and distant, suggesting communication via neuronal networks. The findings suggest that a therapeutic approach targeting all three processes may be needed.