This systematic review aims to (i) evaluate functional near infrared spectroscopy (fNIRS) walking study design in young adults, older adults and people with Parkinson's disease (PD); (ii) examine signal processing techniques to reduce artefacts and physiological noise in fNIRS data; and (iii) provide evidence-based recommendations for fNIRS walking study design and signal analysis techniques. An electronic search was undertaken. The search request detailed the measurement technique, cohort and walking task. Thirty-one of an initial yield of 73 studies satisfied the criteria. Protocols and methods for removing artefacts and noise varied. Differences in fNIRS signals between studies were found in rest vs. walking, speed of walking, usual vs. complex walking and easy vs. difficult tasks. In conclusion, there are considerable technical and methodological challenges in conducting fNIRS studies during walking which can introduce inconsistencies in study findings. We provide recommendations for the construction of robust methodologies and suggest signal processing techniques implementing a theoretical framework accounting for the physiology of haemodynamic responses.