Abstract. Recent studies have provided new evidence that models may systematically underestimate cloud solar absorption compared to observations. This study extends previous work on this "absorption anomaly'' by using observational data together with solar radiative transfer parameterisations to calculate f_s (the ratio of surface and top of the atmosphere net cloud forcings) and its latitudinal variation for a range of cloud types. Principally, it is found that (a) the zonal mean behaviour of f_s varies substantially with cloud type, with the highest values obtained for low clouds; (b) gaseous absorption and scattering can radically alter the pattern of the variation of f_s with latitude, but gaseous effects cannot in general raise f_s to the level of around 1.5 as recently determined; (c) the importance of the gaseous contribution to the atmospheric ASR is such that whilst f_s rises with surface albedo, the net cloud contribution to the atmospheric ASR falls; (d) the assumed form of the degree of cloud overlap in the model can substantially affect the cloud contribution to the atmospheric ASR whilst leaving the parameter f_s largely unaffected; (e) even large uncertainties in the observed optical depths alone cannot account for discrepancies apparent between modelled and newly observed cloud solar absorption. It is concluded that the main source of the anomaly may derive from the considerable uncertainties regarding impure droplet microphysics rather than, or together with, uncertainties in macroscopic quantities. Further, variable surface albedos and gaseous effects may limit the use of contemporaneous satellite and ground-based measurements to infer the cloud solar absorption from the parameter f_s.