Relationships between modeled and measured meteorological state parameters and cloudy and cloud-free conditions are examined using data taken over the ARM (Atmospheric Radiation Measurement) Southern Great Plains Central Facility between 1 March 2000 and 28 February 2001. Cloud vertical layering was determined from the Active Remotely Sensed Cloud Location product based on the ARM active sensor measurements. Both temperature and relative humidity (RH) observations from balloon-borne Vaisala RS80-15LH radiosonde (SONDE) and the Rapid Update Cycle (RUC) 40-km resolution model are highly correlated, but the SONDE RHs generally exceed those from RUC. Inside cloudy layers, the RH from SONDE is 2--14% higher than the RH from RUC at all pressure levels. Although the layer mean RH within clouds is much greater than the layer mean RH outside clouds or in clear skies, RH thresholds chosen as a function of temperature can more accurately diagnose cloud occurrence for either data set than a fixed RH threshold. For overcast clouds (cloud amount greater than or equal to 90%), it was found that the 50% probability RH threshold for diagnosing a cloud, within a given upper tropospheric layer, is roughly 90% for the SONDE and 80% for RUC data. For partial cloud cover (cloud amount is less than 90%), the SONDE RH thresholds are close to those for RUC at a given probability in upper tropospheric layers. Cloud probability was found to be only minimally dependent on vertical velocity. In the upper troposphere, SONDE ice-supersaturated air occurred in 8 and 35% of the clear and cloudy layers, respectively. The RH was distributed exponentially in the ice supersaturated layers as found in previous studies. The occurrence of high-altitude, ice-supersaturated layers in the RUC data was roughly half of that in the SONDE data. Optimal thresholds were derived as functions of temperature to define the best RH thresholds for accurately determining the mean cloud cover. For warm clouds the typical SONDE threshold exceeds 87%, while the RH thresholds for cold clouds are typically less than 80% and greater than 90% with respect to liquid and ice water, respectively. Preliminary comparisons with satellite data suggest that the relationships between cloudiness and RH and T determined here could be useful for improving the characterization of cloud vertical structure from satellite data by providing information about low-level clouds that were obscured by high-level clouds viewed by the satellite. The results have potential for improving computations of atmospheric heating rate profiles and estimates of aircraft icing conditions. Similar analyses are recommended for later versions of the RUC analyses and forecasts.