Dust is an ubiquitous component of the interstellar medium, playing a key role in the energy budget: it absorbs the ultraviolet-to-optical starlight, reemitting the absorbed energy at far-infrared wavelengths. This makes the modeling of dust attenuation crucial to reliably interpret the integrated spectral energy distributions of galaxies in terms of physical parameters, such as mass, star formation rate, star formation history and metallicity. I will start by comparing the results from three sophisticated, popular radiative transfer models of dust attenuation, showing their consistency and some general properties that can be drawn from them. Then, using an original approach, I will show how such models can be used to constrain the content and spatial distribution of dust in a sample of 22 000 nearby star forming galaxies. Finally, I will present a new approach to exploit integrated ultraviolet-to-near infrared colors to constrain the global star formation properties of highly attenuated galaxies, even in the absence of far-infrared observations.