In general circulation models, clouds are parameterized and radiative transfer calculations are performed using the plane-parallel approximation over the cloudy fraction of each model grid. The albedo bias resulting from the plane-parallel representation of spatially heterogeneous clouds has been extensively studied, but the impact of entrainment-mixing processes on cloud microphysics has been neglected up to now. In this paper, this issue is examined by using large-eddy simulations of stratocumulus clouds and tridimensional calculations of radiative transfer in the visible and near-infrared ranges. Two extreme scenarios of mixing are tested: the homogeneous mixing scheme with constant concentration and reduced droplet sizes, against the inhomogeneous mixing scheme, with reduced concentration and constant droplet sizes. The tests reveal that entrainment-mixing effects at cloud top may substantially bias the simulated albedo. In the worse case, which corresponds to a fragmented and thin stratocumulus cloud, the albedo bias changes from −3% to −31% when using both mixing schemes alternatively.