The reflectance-based approach is a calibration method in which in-situ radiometric measurements of a selected target on the Earth surface and atmospheric parameters are used as input into a radiative transfer model for estimating a top-of-atmosphere (TOA) radiance. However, an estimate of the TOA radiance is incomplete unless complemented with its uncertainty. This work described a methodology for predicting TOA radiance with an associated uncertainty. Two different methods of uncertainty propagation were applied: ISO GUM and Monte Carlo. In both methods the TOA radiance was calculated using the atmospheric radiative transfer code MODTRAN. Vicarious calibration data collected on August 21st, 2014 at Atacama Desert, Chile and on July 25th, 2014 at Luís Eduardo Magalhães, Brazil were used to evaluate the methodology. The results, for the wavelength between 350 to 2400nm, showed that using an agricultural area covered by soil surface (Brazil) the final uncertainty of the TOA radiance ranged from 1.0% to 9.2%; and using a desert surface (Chile) the final uncertainty ranged from 0.7% to 4.5%. The results also showed that the overall uncertainty using Monte Carlo simulations were smaller than the root sum of squares of each parameter (ISO GUM), suggesting that there is correlation among some parameter that is not take into account when the data are considered independent.