Masonry building aggregates can be found throughout historical city centers of Europe. Adjacent buildings have often developed during the long densification process, without consistent planning or engineering. It is common for neighboring units to share a structural wall with the connection between the units ensured through interlocking stones or a dry joint. Another consequence of the gradual development through the years are the different material properties, distributions of openings, and different roof and floor heights of the neighboring units. The difference in properties and the uncertainty related to the connection between the units leads to difficulties when modelling the seismic response. For that reason, units of an aggregate are often modelled as separate or perfectly connected. Both simplifications can lead to an incorrect estimation of the aggregate response. One of the principal reasons impeding the advancements in this area is the lack of large-scale experimental campaigns, due to the size and the cost of such campaigns. For that reason, the project AIMS (Seismic Testing of Adjacent Interacting Masonry Structures), included in the H2020 project SERA has the objective to provide the experimental data on the behaviour of an aggregate of two stone masonry buildings under bi-directional horizontal acceleration. The test unit is constructed at half scale, with one storey building and a two-storey building. The buildings share one common wall and the façade walls are connected by dry joints. Units have different floor heights and floor beam orientations, leading to a complex dynamic response. The shake table test is conducted at the LNEC seismic testing facility. Extensive instrumentation, including an optical displacement measurement system, displacement transducers, and accelerometers are used to provide information on the force-displacement response. Special attention is devoted to the interface opening and the global behaviour in relation to the interface separation.