The new Heliosat-4 method estimates the downwelling shortwave irradiance received at ground level in all sky conditions. It provides the global irradiance and its direct and diffuse components on a horizontal plane and the direct irradiance for a plane normal to sun rays. It is a fully physical model using a fast, but still accurate approximation of radiative transfer modelling and is therefore well suited for geostationary satellite retrievals. It can also be used as a fast radiative transfer model in numerical weather prediction models. It is composed of two models based on abaci, also called look-up tables: the already-published McClear model calculating the irradiance under cloud-free conditions and the new McCloud model calculating the extinction of irradiance due to clouds. Both have been realized by using the libRadtran radiative transfer model. The main inputs to Heliosat-4 are aerosol properties, total column water vapour and ozone content as provided by the Copernicus Atmosphere Monitoring Service (CAMS) every 3 h. Cloud properties are derived from images of the Meteosat Second Generation (MSG) satellites in their 15 min temporal resolution using an adapted APOLLO (AVHRR Processing scheme Over cLouds, Land and Ocean) scheme. The 15 min means of irradiance estimated by Heliosat-4 are compared to corresponding measurements made at 13 stations within the Baseline Surface Radiation Network and being located in the field of view of MSG and in various climates. The bias for global irradiance is comprised between 2 and 32 W m −2. The root mean square error (RMSE) ranges between 74 and 94 W m −2. Relative RMSE values range between 15 % and 20 % of the mean observed irradiance for stations in desert and Mediterranean climates, and between 26 % and 43 % for rainy climates with mild winters. Correlation coefficients between 0.91 and 0.97 are found. The bias for the direct irradiance at normal incidence is comprised between −163 and +50 W m −2. The RMSE ranges from 160 W m −2 (29 % of the mean observed irradiance) to 288 W m −2 (63 %). The correlation coefficient ranges between 0.67 and 0.87.