Abstract In evaluating the β-spectrum of atomic 3H, a more accurate formula is obtained, in which the first three transitions leading to the lowest three states of 3He+ are treated exactly, and the closure approximation is used to estimate the contribution from all excited states with n > 3. The formula thus obtained is compared with that used by Lyubimov et al. in determining the neutrino mass. As a result, it has been shown that the formula used by Lyubimov et al. does not yield an adequate accuracy. Especially the end-point energy they obtained may differ from the true value by 13.6 eV, which would cause an uncertainty of the same order in the value of the neutrino mass.