A numerical model to estimate spectral and broadband ultraviolet irradiance (290 to 325 nm) for Canadian conditions is described and validated with Brewer spectroradiometer measurements at four stations. The model applies the delta-Eddington algo rithm to a 50-level, 100-km, plane-parallel atmosphere with cloud inserted between 2 and 3 km. It requires measured total atmospheric ozone depth and hourly observations of cloud amount. In the absence of ozone soundings, model ozone profiles are scaled by the ratio of measured (from the Brewer instrument) to model total atmospheric ozone depths. In the model calculations, SUSIM ATLAS 3 extraterrestrial irradiance measurements are averaged for each nanometer of wavelength to mimic the triangular filter used by the Brewer instrument. Ozone absorption is calculated from the temperature-dependent coef ficients of Pauer and Bass (1985), Rayleigh optical depths after Elterman (1968), and aero sol optical properties from MODTRAN. Surface albedo is a function of snow depth and 0.05 for snow-free ground. Model and measured spectral irradiances for cloudless skies agree well, but model values are smaller than measurements for wavelengths below about 305 nm because of enhancement of the Brewer signal by stray light. Model values of daily cloudless sky irradiance using lidar aerosol optical depth measurements from York Univer sity after the Mt. Pinatubo eruption in 1991 agree well with measurements. Cloud optical depths were calculated iteratively for overcast conditions. A fixed optical depth of 45 was used to calculate cloudy sky irradiances at the four stations. These agree well with mea surements. Mean bias error (MBE) is less than 5% of the mean measured daily irradiance and root mean square error (RMSE) less than 25%, decreasing to below 12% for 10-day averages. Agreement between mean daily measured and calculated spectral irradiances over a month is also good. [