Shape memory polymers (SMPs) belong to an emerging and fast developing branch of smart materials. SMPs can fix temporary shapes and recover their original one upon exposure to external stimuli. According to the latter SMPs can be grouped in thermal-, electro-, magnetic-, light-, solution (water)-activated versions. Other categorizations consider the chemical structure, the number of shapes SMP can memorize in one cycle and the possibility to repeat temporary/original shaping just upon external stimulus [1][2][3]. Nowadays, exhaustive reviews are available on the structure, properties and applications of SMPs [2][3][4][5]. SMPs consist of permanent netpoints and molecular switches of reversible nature. The latter are exploited for setting the temporary shape whereas the former memorize the original shape. For cured thermosets the netpoints are given by covalent crosslinks of the chemical network. This network architecture enables keeping the stable (original) shape also after recovery. The segments between the crosslinks act as molecular switches. They undergo a reversible phase transition at the glass transition temperature (T g ) range upon heating, where heating is the external stimulus. The transition between the glassy and rubbery states is associated with a large drop in the modulus of SMPs (covering almost 3 orders of magnitude). Temporary shaping, performed in the rubbery state, yields a conformational rearrangement in the segments of the chemically crosslinked structure. This causes a prominent entropy reduction. The shape is fixed upon cooling to the glassy state whereby energy is stored via 'freezing in' the related network deformation. When the material is heated above the T g again, the stored energy is released by Abstract. In this study asymmetrically reinforced epoxy (EP)/carbon fibre (CF) fabric composites were prepared and their shape memory properties were quantified in both unconstrained and fully constrained flexural tests performed in a dynamic mechanical analyser (DMA). Asymmetric layering was achieved by incorporating two and four CF fabric layers whereby setting a resin-and reinforcement-rich layer ratio of 1/4 and 1/2, respectively. The recovery stress was markedly increased with increasing CF content. The related stress was always higher when the CF-rich layer experienced tension load locally. Specimens with CF-rich layers on the tension side yielded better shape fixity ratio, than those with reinforcement layering on the compression side. Cyclic unconstrained shape memory tests were also run up to five cycles on specimens having the CF-rich layer under local tension. This resulted in marginal changes in the shape fixity and recovery ratios.