AbstractASTRI-Horn is an Imaging Atmospheric Cherenkov Telescope characterized by a dual-mirror optical system with a primary mirror diameter of 4.3 m and a curved focal surface covered by silicon photomultiplier (SiPM) sensors managed by an innovative fast front-end electronics. ASTRI-Horn is installed in Italy at the INAF “M.C. Fracastoro” observing station (Mount Etna, Italy); it is the prototype of nine similar telescopes forming the ASTRI MiniArray that will be installed at the Teide Astronomical Observatory, in Tenerife (Canary Islands, Spain). In the ASTRI-Horn camera, the output signals from SiPMs are AC coupled to the front-end electronics stopping any slow varying signals. However, the random arrival of the night sky background photons produces fast fluctuations in the signal that the electronics is able to detect. The noise generated by this effect is proportional to the level of the diffuse night sky background. In this work, we present the analysis of the background data in ASTRI-Horn observations during the period December 2018–March 2019, using images of triggered showers. We compare the results relative to 2018 December 7-8 and 2019 March 6-7 nights with the contemporary night sky background fluxes measured by UVscope. This is a small auxiliary instrument mounted on the external structure of the ASTRI-Horn telescope and devoted to the night sky background evaluation in the UV band. A strong correlation between the considered data was detected. This correlation can be a diagnostic tool to assure the proper behavior of the ASTRI-Horn camera in view of the ASTRI MiniArray implementation. ASTRI-Horn is also equipped with the Variance technique able to sample the level of the pixel signals in absence of showers with an high rate. The method presented in this paper, based on shower images, is a new approach that has never been investigated until now. It does not substitute the Variance, that will the baseline for the background evaluation after exhaustive testings, but it is complementary to it when Variance data are available. This is the only one method working very well, that can be applied whenever the standard Variance method is not operative.