The increasing requirement for light weight constructions and the unsatisfactory performances of traditional metals and conventional engineering materials especially their deficiencies in positively respond to environmental stimuli, have made the necessity for development of alternative materials. Such alternative materials are the so-called adaptive, multifunctional, smart or intelligent composites to facilitate the realization of some engineering applications that are difficult to achieve with the existing conventional materials. Composite materials have found increasing applications in construction, aerospace and automotive industries due to their good characteristics of light weight, improved strength, corrosion resistance, controlled anisotropic properties, and reduced manufacturing and maintenance costs. However, there is a growing demand to improve on composite materials to have “smart" capabilities so as to be able to sense, actuate and respond to the surrounding environment. Shape memory alloys (SMAs) are metallic alloys that can undergo martensitic phase transformations as a result of applied thermomechanical loads and are capable of recovering permanent strains when heated above a certain transformation temperature. SMAs possess sensing and actuating functions and have the potential to control the mechanical properties and responses of their hosts due to their inherent unique characteristics: shape memory effect (SME) and pseudoelasticity (PE). When integrated into structural components, they perform sensing, diagnosing, actuating and repair or healing functions, thereby enhancing improved performance characteristics of their hosts. Amongst the commercially available SMAs, NiTi (Nickel Titanium) in the forms of wires, ribbons, bars, particles and porous bulks are the most widely used one because of its excellent mechanical properties and superior material characteristics. Embedding SMAs into composite materials can create smart or intelligent hybridized composites. The success of conventional fiber reinforced composites (FRC) relies on the quality of bonding between fibers and matrix. The excellent interfacial bond strength between shape memory alloy (SMA) inclusions and the host materials is also critically important for success of SMA-composites. A literature review shows that there is a lack of simulation models on the interfacial behaviours of SMA composites. Therefore, in the past, the operation limit as well as the ideal actuation condition of SMA inclusions could not be predicted during the design step. The simulation models in this research provide a study basis for the prediction of internal stresses and interfacial strength of the SMA composites. In this research has been tried to investigate stresses in SMA composites. Therefore, at first equations that describe SMA behaviours have been explained. Afterward according to described equations a subroutine in a finite element package (Abaqus) has been written to introduce SMA materials to it. This subroutine has been written in two ways; first one-dimensional subroutine and then three-dimensional subroutine for SMA materials have been provided. Thus, too many models in different conditions such as different lengths and temperatures have been simulated. According to these simulations, different stress distributions in different situations have been caught. Afterward the failure in this material has been investigated. For this purpose, the cohesive element has been used and the stress in which partial debonding in fiber and matrix would happen, has been investigated. Some results of these simulations are diagrams of debonding stress according to temperature in different lengths. According to these diagrams and simulations, designers can easily find out the best conditions for SMA-composites in which debonding would happen sooner or later. These diagrams also show that debonding stress does not depend on length of embedded fiber to matrix. Keyword : Composite, Shape memory alloy, Interfacial stress, Cohesive element, Debonding stress.