Abstract. The limb-scatter satellite viewing geometry is well suited to detecting low-concentration aerosols in the upper troposphere due to its long observation path length (~200 km), high vertical resolution (~1–2 km) and good geographic coverage. We use the fully three-dimensional radiative transfer code SASKTRAN to simulate the sensitivity of limb-scatter viewing Odin/OSIRIS satellite measurements to absorbing mineral dust and carbonaceous aerosols (smoke and pure soot), as well as to non-absorbing sulfate aerosols and ice in the upper troposphere. At long wavelengths (813 nm) the addition of all aerosols (except soot) to an air only atmosphere produced a radiance increase as compared to air only, on account of the low Rayleigh scattering in air only at 813 nm. The radiance reduction due to soot aerosol was negligible (<0.1%) at all heights (0–100 km). At short wavelengths (337, 377, 452 nm), we found that the addition of any aerosol species to an air only atmosphere caused a decrease in single-scattered radiation due to an extinction of Rayleigh scattering in the direction of OSIRIS. The reduction was clearly related to particle size first, with absorption responsible for second-order effects only. Multiple-scattered radiation could either increase or decrease in the presence of an aerosol species, depending both on particle size and absorption. Large scatterers (ice, mineral dust) all increased multiple-scattered radiation within, below and above the aerosol layer. Small, highly absorbing pure soot particles produced a negligible multiple-scattering response (<0.1%) at all heights, primarily confined to within and below the soot layer. Medium-sized scatterers produced a multiple-scattering response that depended on their absorbing properties. Increased radiances were simulated as compared to air only at all short wavelengths (337, 377 and 452 nm) for sulfate aerosol particles (non-absorbing) while decreased radiances were simulated for smoke particles (absorbing) at 337 and 377 nm, where multiple scattering involving the medium-sized carbonaceous particles amplified their absorbing properties. At 452 nm, however, this effect was attenuated and albedo-dependent. At short wavelengths, the combined effect of single scattering decreases and multiple scattering increases led to complex total radiance signatures that generally could not unambiguously distinguish absorbing versus non-absorbing aerosols. Smoke aerosols led to a total radiance decrease (as compared to air only) at all altitudes above the aerosol layer (15–100 km). This unique signature was a result of the aerosols' strong absorbing properties, non-negligible scattering efficiency, and number concentrations high enough to make multiple scattering effects due to the aerosol itself significant. Thus, in the limb-scatter viewing geometry scene darkening above the aerosol layer is unambiguously due to absorption whereas scene darkening within and below the aerosol layer can simply be the result of a reduction in single-scattered radiance. Our simulations show a greater scene darkening for decreasing wavelengths, increasing surface albedo, decreasing solar zenith angle, and increasing particle number concentration, however, at 337 nm this effect did not exceed 0.5% of the total radiance due to air only, making the unique identification of medium-sized carbonaceous aerosols, i.e., smoke, difficult. Scene darkening (or brightening) varies linearly with particle number concentration over three orders of magnitude. A fortuitous, unexpected implication of our analysis is that limb-scatter retrievals of aerosol extinction are not sensitive to external information about aerosol absorption.