Geomorphological evolution, erosion and retreat processes that affect the rocky coasts of the mid-western Adriatic Sea (Abruzzo, Central Italy) are the subject of this research. This coastal sector, one of the few examples of clastic soft rock coasts in the Mediterranean Sea, is characterized by active, inactive and paleo cliffs, as well as coastal slopes, composed of the clayey-sandy-arenaceous-conglomeratic marine sequence (Early-Middle Pleistocene) covered by continental deposits (Late Pleistocene-Holocene). This study provides geomorphological and 3D modelling stability analyses of the cliffs of Torre Mucchia, Punta Lunga, Punta Ferruccio (Ortona, CH) and Punta Aderci (Vasto, CH), which are popular tourist sites included in natural reserve areas. They are representative of two main types of active cliffs on soft clastic rocks: cliffs on sandstone and cliffs on conglomerate with notches. In order to evaluate the processes and factors that induce cliffs to retreat and their recent evolution, the research was based on a DEM analysis (LIDAR 2 × 2 m data), aerial photos and an orthoimages interpretation, detailed geological–geomorphological surveys, and a structural analysis; field and remote investigations were combined with numerical modelling with a FLAC3D calculation code. Geological and geomorphological field data provided reliable 3D models, and FLAC3D numerical analyses allowed the definition of the most critical and/or failure areas, and the evaluation of the controlling factors, evolution mechanisms of the slopes and the sliding kinematics of gravitational instability phenomena. Different retreat mechanisms have been observed all along the investigated coastal sectors, induced by gravitational processes due to coastal erosion cycles at the foot of the cliffs, and controlled by lithological features and joints systems. The geomorphological analysis combined with the 3D modelling (i) showed that the retreat process of the cliffs is connected to translational slides and rockfalls (cliffs on sandstone), combined rockfalls, and topples (cliffs on conglomerate), largely controlled by main joints; (ii) defined the most critical areas along the cliffs. These results are of great interest in the assessment of hazard connected to potential sliding on the cliffs. Their implementation within Geographical Information Systems provides a valuable contribution to the integrated management of coastal areas, strongly improving the identification and prediction of landscape changes and supporting a new geomorphological hazards assessment, in areas of high tourism, as well as natural and cultural landscape value.