Thermally activated shape memory polymers are typically programmed by initially heating the material above the glass transition temperature (Tg), deforming to the desired shape, cooling below Tg, and unloading to fix the temporary shape. This process of deforming at high temperatures becomes a time-, labor-, and energy-expensive process while applying to large structures. Alternatively, materials with reversible plasticity shape memory property can be programmed at temperatures well below the glass transition temperature which offers several advantages over conventional programming. Here, the free, partial, and fully constrained recovery analysis of cold-programmed multi-walled carbon nanotube–reinforced epoxy nanocomposites is presented. The free recovery analysis involves heating the temporary shape above Tg without any constraints (zero stress), and for fully constrained recovery analysis, the temporary shape is held constant while heating. The partially constrained recovery behavior is studied by applying a constant stress of 10%, 25%, and 50% of the maximum recovery stress obtained from the completely constrained recovery analysis. The samples are also characterized for their thermal, morphological, and mechanical properties. A non-contact optical strain measurement method is used to measure the strains during cold-programming and shape recovery. The different recovery behaviors are analyzed by using a thermo-viscoelastic–viscoplastic model, and the predictions are compared with the experimental results. © 2018, © The Author(s) 2018.