Most standard air temperature measurements are subject to significant errors mainly due to sensor heating by solar radiation, even when the measurement principle is accurate and precise. We present various air temperature measurements together with other measurements of meteorological parameters using different sensor systems at a snow-covered and a vegetated site. Measurements from naturally ventilated air temperature sensors in multiplate shields are compared to temperatures measured using sonic anemometers which are unaffected by solar radiation. Over snow, 30 min mean temperature differences can be as large as 10°C. Unshielded thermocouples were also tested and are generally less affected by shortwave radiation. Temperature errors decrease with decreasing solar radiation and increasing wind speed but do not completely disappear at a given solar radiation even in the presence of effective ventilation. We show that temperature errors grow faster for reflected than for incident solar radiation, demonstrating the influence of the surface properties on radiative errors, and we detect the albedo as a variable with major influence on the magnitude of the error as well as a key quantity in possible error correction schemes. An extension is proposed for an existing similarity regression model to correct for radiative errors; thus, surface-reflected shortwave radiation is identified as a principal source of error and the key variable for obtaining a unique nondimensional scaling of radiative errors.