An analytical approach for the calculation of radiative self-absorption in flames is presented, with the objective of providing an effective and computationally inexpensive tool for selecting radiation modeling methods for combustion applications. This one-dimensional non-local model approximates the flame structure as an infinitely long cylinder where all properties vary only along the radial direction. Flamelet solutions are then mapped to the radial direction and the radiative source terms, in particular the absorption, are computed analytically. The model is validated for a laminar jet flame and a turbulent pool fire, with results for the net radiative loss and the flame self-absorption showing good agreement to those of multidimensional, coupled simulations. To illustrate its applicability, the model is employed to study the characteristics of radiation in high-pressure combustion, in fire suppression and in hydrogen combustion. For the pressurized case, increased significance of radiation self-absorption is observed compared to an atmospheric flame, and accounting for the non-gray nature of radiation is found to be critical. In the fire suppression case, comparisons of the S-curves show that the choice of radiation model modifies the lower extinction limit, which becomes more important with reduced oxygen content. Lastly, for the hydrogen flame, the Planck-mean absorption coefficient is found to be more uniformly distributed in the mixture fraction space, and the optical thickness is smaller than a hydrocarbon flame at comparable conditions, indicating that radiation modeling requirements for that flame are likely less stringent. In general, the findings regarding radiation characteristics made on the basis of the non-local model are consistent with observations reported in the literature, but without the need for complex coupled combustion-radiation calculations. Therefore, the model provides a viable approach for selecting appropriate radiation models in combustion simulations.