Neurofibrillary tangles, one of the hallmarks of Alzheimer's disease, had been a target of modern neuropathology based on electron microscopy. In 1960s their unit fibrils were found to be paired helical filaments (PHF), the unique appearance of which attracted many researchers to their nature. In the late 1970s, a keen interest in their constituents at the molecular levels had increasingly grown, but electron microscopic approach failed to address the issue. I describe here what was going on at the turning point when electron microscopic study yielded immunocytochemical approach and direct characterization by isolation, with some emphasis on the situation in Japan. Personal memories are provided about Dr Selkoe's lab (1981-1982) in Mailman Research Center, Belmont, Boston, where we encountered a series of remarkable properties of PHF. How insolubility of PHF, and smearing on the blot was found is described.