Microglia play a critical role in many processes fundamental to learning and memory in health and are implicated in Alzheimer's pathogenesis. Minocycline, a centrally-penetrant tetracycline antibiotic, inhibits microglial activation and enhances long-term potentiation, synaptic plasticity, neurogenesis and hippocampal-dependent spatial memory in rodents, leading to clinical trials in human neurodegenerative diseases. However, the effects of minocycline on human memory have not previously been investigated. Utilising a double-blind, randomised crossover study design, we recruited 20 healthy male participants (mean 24.6 ± 5.0 years) who were each tested in two experimental sessions: once after 3 days of Minocycline 150 mg (twice daily), and once 3 days of placebo (identical administration). During each session, all completed an fMRI task designed to tap boundary- and landmark-based navigation (thought to rely on hippocampal and striatal learning mechanisms respectively). Given the rodent literature, we hypothesised that minocycline would selectively modulate hippocampal learning. In line with this, minocycline biased use of boundary- compared to landmark-based information (t980 = 3.140, p = 0.002). However, though this marginally improved performance for boundary-based objects (t980 = 1.972, p = 0.049), it was outweighed by impaired landmark-based navigation (t980 = 6.374, p < 0.001) resulting in an overall performance decrease (t980 = 3.295, p = 0.001). Furthermore, against expectations, minocycline significantly reduced activity during memory encoding in the right caudate (t977 = 2.992, p = 0.003) and five other cortical regions, with no significant effect in the hippocampus. In summary, minocycline impaired human spatial memory performance, likely through disruption of striatal processing resulting in greater biasing towards reliance on boundary-based navigation.