The radiative characteristics of low-level clouds over the Arctic ice pack in spring are inferred from ground-based broadband flux measurements using a radiative transfer model. An informal comparison of several radiative transfer models is performed for clear-sky conditions observed in April during LEADEX. The broadband longwave and shortwave surface fluxes obtained with Streamer and the broadband longwave surface flux obtained with the Rapid Radiative Transfer Model (RRTM) correlate well with the observations. The Streamer code is chosen to determine cloud optical and microphysical properties based on its performance under clear-sky conditions, its sophisticated treatment of clouds, and its option for calculating fluxes using a discrete ordinate technique. The radiative properties of the Arctic atmosphere are complicated by the presence of clouds. Low clouds occurred 30% of the time during LEADEX. The longwave and shortwave optical depths of clouds based below 2 km are inferred from surface-based measurements of cloud base height and broadband radiative fluxes by matching the observed fluxes with those obtained with the four-stream radiative transfer code available as part of Streamer. The retrieved broadband shortwave and longwave optical depths for low clouds observed during LEADEX ranged between 0.2 and 3. An attempt is made to infer cloud phase and cloud particle size from the ratio of the retrieved broadband shortwave and longwave cloud optical depths. This ratio ranges from 1.33 to 1.75 for ice, from 0.85 to 1.9 for liquid, and from 0.85 to 1.9 for mixed-phase clouds. Although there is much overlap between the ranges, it is shown that liquid-water clouds are characterized by optical depth ratios that are generally smaller than those found in ice clouds. This method for inferring cloud microphysics from surface-based measurements of broadband radiative fluxes and cloud base height may prove useful where more sophisticated observations of cloud microphysics are lacking. ¿ 1997 American Geophysical Union