Disclosed are a method and apparatus for accurately measuring nuclear magnetic resonance (NMR) longitudinal axis relaxation time by minimizing the influence of the amount of blood. The method includes applying a first magnetic field M.sub.O to a living body portion into which the blood flows magnetizing the blood in a first direction; applying a second magnetic field M.sub.Z to at least a portion of the living body portion, to which the first magnetic field M.sub.O is applied, magnetizing at least a portion of the blood in a second direction vertical to the first direction;; blocking M.sub.Z detecting a signal that is induced as magnetization is returned from the second direction to the first direction; detecting the amount of blood to which the second magnetic field M.sub.Z is applied; and calculating the longitudinal axis relaxation time T1 of the blood based on the induced signal and based on the amount of blood.