In software, faults can occur and may disturb the normal behavior or working of a system. Software fault tolerance techniques must be implemented to remove these faults. Fault tolerance is defined as how the system fights when faults occur. A system cannot be truly fault tolerant until software fault tolerance techniques are applied to it. In early research, it is estimated that almost 60-90% of system failures are attributed to software failures. These failures can be controlled by applying software fault tolerance techniques. In this research, a comparative study of different software fault tolerance techniques (like Swift, Trump, Mask and N-version programming) will be carried out and current research challenges in these techniques will be discussed.