This study examines the accuracy of the model 501A UV biometer in taking measurements relative to a predetermined background condition. The UVSPEC model with the discrete ordinates radiative transfer (DISORT) algorithm was used to mimic a range of radiation environments covering differing zenith angles, aerosol, and cloud regimes. These data were used to model typical responses for two hypothetical biometers with differing spectral and cosine responses. A ratio was then calculated between the irradiance for any zenith angle, atmospheric opacity, and ozone load and the irradiance for the same zenith angle and ozone load but for a clean atmosphere with 95-km visibility conditions. Both virtual instruments gave very similar ratios for a wide variety of aerosol loads, cloud cover, and zenith angles. Furthermore, the instrument ratios also agreed with ratios predicted by an instrument with a perfect cosine response and responding to the erythema curve. The above theoretical results were tested with three model 501A biometers that were sampling the same environment for a three-month period in Hobart, Australia. Ratios were calculated for each instrument between any 10-min measured average irradiance and the instrument value predicted for the same ozone load and zenith angle but for cloudless conditions. Instrument ratios for solar zenith angles less than 70¿ agree closely with each other, with a coefficient of variance exceeding 0.99 and RMS errors of between 1 and 2%. Finally, the data may be turned into absolute corrected irradiance by using the UVSPEC calculations as discussed in the paper.