Alzheimer's disease pathology is characterized by extracellular deposits of amyloid-␤ (A␤) and intracellular inclusions of hyperphosphorylated tau. Although genetic studies of familial Alzheimer's disease suggest a causal link between A␤ and disease symptoms, the failure of various A␤-targeted strategies to slow or halt disease progression has led to consideration of the idea that inhibition of tau aggregation might be a more promising therapeutic approach. Methylene blue (MB), which inhibits tau aggregation and rescue memory deficits in a mouse model of tauopathy, however, lacked efficacy in a recent Phase III clinical trial. In order to gain insight into this failure, the present study was designed to examine the mechanism through which MB inhibits tau aggregation. We found that MB inhibits heparin-induced tau aggregation in vitro, as measured by thioflavin T fluorescence. Further, MB reduced the amount of tau in precipitants recovered after ultracentrifugation of the aggregation mixture. Atomic force microscopy revealed that MB reduces the number of tau fibrils but increases the number of granular tau oligomers. The latter result was confirmed by sucrose gradient centrifugation: MB treatment was associated with higher levels of granular tau oligomers (fraction 3) and lower levels of tau fibrils (fractions 5 and 6). We previously demonstrated that the formation of granular tau oligomers, rather than tau fibrils, is essential for neuronal death. Thus, the fact that MB actions are limited to inhibition of tau fibril formation provides a mechanistic explanation for the poor performance of MB in the recent Phase III clinical trial.