A modified version of the Monkman-Grant equation, which can provide a more accurate means of predicting creep rupture life than the standard Monkman-Grant formula, has been investigated for Gr.91 using creep data in the NIMS Creep Data Sheets at 450–725 °C. The maximum time to rupture tr was 1.2 × 105 h. The tr versus minimum creep rate ε˙min plot, which is called the Monkman-Grant relation, deviates downward at low stresses and long times. The difference between the maximum and minimum tr at a same ε˙min becomes more significant with decreasing stress and increasing test duration. Better correlation of the tr with the ε˙min is not obtained by the replacement of tr with (tr/εr), where εr is the total or rupture strain. The magnitude of data scattering is larger in the (tr/εr) versus ε˙min plot than in the tr versus ε˙min plot. The (tr/εm), where εm is the strain to minimum creep rate, is inversely proportional to the ε˙min over a wide range of stress, temperature and test duration and the magnitude of data scattering is only a little bit even at low stresses and long times. The (tr/εm) versus ε˙min plot gives us more reliable relation for Gr.91 than the tr versus ε˙min and (tr/εr) versus ε˙min plots.