The applicability of biased and random error models for determining the uncertainty of photometer quality factor using Monte Carlo simulations is investigated. In the biased error model the contribution to the measured value has the same sign and almost the same magnitude at neighbouring wavelengths, whereas in the random error model contributions of varying signs appear at the neighbouring wavelengths. Both error models are used with real spectral responsivity data and uncertainty budgets of two photometers. Analytical considerations supported by simulation results show that the random error model alone may easily underestimate the uncertainty of . The biased error model is recommended as the basis of the uncertainty evaluation of . It is also demonstrated by real photometer data that the uncertainty of can be highly sensitive to fine details of the photometer spectral responsivity when the biased error model is used.