Methods of obtaining stellar luminosities (L) have been revised and a new concept, standard stellar luminosity, has been defined. In this paper, we study three methods: (i) a direct method from radii and effective temperatures; (ii) a method using a mass-luminosity relation (MLR); and (iii) a method requiring a bolometric correction. If the unique bolometric correction (BC) of a star extracted from a flux ratio (fv/f(Bol)) obtained from the observed spectrum with sufficient spectral coverage and resolution are used, the third method is estimated to provide an uncertainty (Delta L/L) typically at a low percentage, which could be as accurate as 1 per cent, perhaps more. The typical and limiting uncertainties of the predicted L of the three methods were compared. The secondary methods, which require either a pre-determined non-unique BC or MLR, were found to provide less accurate luminosities than the direct method, which could provide stellar luminosities with a typical accuracy of 8.2-12.2 per cent while its estimated limiting accuracy is 2.5 per cent.