Outgoing longwave (LW) radiation (OLR) and infrared heating rate errors due to the neglect of multiple scattering by clouds and horizontal cloud inhomogeneity in infrared radiative transfer parameterizations are assessed with a modified version of the Fu and Liou <1992> radiative transfer model and field measurements of optically thin cirrus that were observed during the fall 1997 and spring 2000 intensive observation periods conducted at the Atmospheric Radiation Measurement Southern Great Plains site in Oklahoma. A rigorous method is first introduced for incorporating atmospheric and cloud optical and geometric properties derived from measurements in radiative transfer calculations and for ensuring the consistency of these calculations with observations. Part of that method is the use of optical depth derived from measurements of a Multi Filter Rotating Shadowband Radiometer (MFRSR) at 415 nm to determine the cloud ice water content for the longwave and shortwave radiative transfer calculations. The optical depth is corrected to account for forward scattering by cirrus crystals, which is dominant under the thin cirrus conditions observed. Surface downwelling fluxes from the radiative transfer calculations are compared against observations from the Baseline Surface Radiation Network for consistency, with primary emphasis placed on achieving good agreement between the observed and calculated direct normal components. For all cases considered, there is very good agreement in the solar direct irradiance and reasonably good agreement in the solar diffuse irradiance, but without the forward scattering correction the agreement for both would degrade; the diffuse agreement, in particular, would do so by 12 W m-2 (>30%). Very good agreement is achieved in the LW with an absolute mean difference of 0.02 W m-2. For the observed cases considered, neglecting multiple scattering of LW by thin cirrus overestimates OLR by 6--8 W m-2 or less (depending on cloud optical depth and particle size) because of exclusion of reflection of upwelling LW at the cloud base and results in heating rate errors of as much as 0.2 K d-1. The effect of horizontally homogeneous clouds is approximated by employing the half hourly mean of the derived optical and geometric properties in the calculation. The impact of the combined effect on OLR and LW heating rates is shown from these calculations to be as high as 30 to 35 W m-2, and heating rate errors are on the order of 0.5 to over 1 K d-1, which are very significant errors.