Rubble stone masonry is a common construction typology of historical city centres and vernacular architecture. While past earthquakes have shown that it is one of the most vulnerable masonry construction typologies, there are few experimental campaigns giving quantitative information on the strength and displacement capacity, as well as on the damage pattern under in-plane horizontal loading. Additionally, there is a lack of experimental studies looking into the role of the size of the structural element on the structural response of rubble stone masonry walls. This paper presents an experimental campaign for testing the in-plane seismic response of rubble stone masonry walls and developing digital twins for their numerical simulation. Nine rectangular walls of three different sizes were constructed by experienced masonry using irregular limestone units and lime mortar. The mechanical properties on the unit and wall level were obtained through mechanical characterization tests on mortar samples and wallets. The walls were tested at the laboratory of EPFL (Switzerland) under quasi-static loading using the same load and boundary conditions. All walls were tested up to axial load failure (i.e. up to the point where the walls cannot bear any vertical load), providing information on the drift levels at collapse of the wall. Digital image correlation measurements were used to obtain displacement fields and extract crack patterns during the load history. The results of this experimental campaign offer a first insight on the size effect on the in-plane response of rubble stone masonry walls. For the purpose of validating numerical modelling techniques for stone masonry structures, we developed the geometrical digital twins of three of the tested walls. Each geometrical digital twin represents a numerical replica of the physical specimen, generated with a pipeline based on laser scanning. In this way, the experimental and numerical datasets are valuable for the calibration of numerical models for rubble stone masonry and for further numerical investigations of the size effect.