The coefficients of linear regression lines fit to hourly observations of atmospheric transmittance and TOA albedo have been used previously to address the problem of anomalous cloud absorption (ACA). While these coefficients indicate, reasonably well, the impact of clouds on atmospheric absorptance for 1D model data, this is not necessarily the case for hourly data. This is because for non-uniform clouds, regression coefficients are reduced significantly relative to 1D model results due to the effects of horizontal transport of radiation and poor sampling of transmittance. These reductions are only illusions of ACA because domain-averaged atmospheric absorptances are almost insensitive to cloud geometry. The ramifications of these effects can be seen in hourly data that were used in a previous study which claimed to support ACA. They are also demonstrated with a 3D Monte Carlo, broadband radiative transfer algorithm acting on data generated by a cloud-resolving model.¿ 1997 American Geophysical Union