Despite the progress made in face and fingerprint, in some forensic scenarios it is not possible to successfully acquire such biometric characteristics. Therefore, the need for other ways of performing biometric recognition is of utmost importance to the research community. Hand anatomy is the key to determine the individuality of hand-based biometrics. The thesis investigated the performance of hand-based biometric recognition systems in forensic investigation scenarios. Three state-of-the-art systems were selected and evaluated over several experiments. In addition, three hand image datasets of varying complexity were used to evaluate the accuracy of biometric recognition in conditions ranging from ideal to challenging. The results presented in this Thesis showed that, while the tested systems can operate reliably in controlled data, their performance was significantly worse in uncontrolled scenarios. The experiments conducted in this thesis indicated that hand position and rotation, as well as image background, can significantly affect the accuracy of the tested models. This report concluded with a proposal for further studies that could counteract these factors and other future research perspectives.