Each year, 75-100 unprovoked shark attacks on humans are recorded, most of them resulting in no or minor injuries, while a few are fatal. Often, shark identification responsible for attacks relies on visual observations or bite wound characteristics, which limits species determination and preclude individual identification. Here, we provide two genetic approaches to reliably identify species and/or individuals involved in shark attacks on humans based on a non-invasive DNA sampling (i.e. DNA traces present on bite wounds on victims), depending on the knowledge of previous attack history at the site. The first approach uses barcoding techniques allowing species identification without any a priori, while the second relies on microsatellite genotyping, allowing species identification confirmation and individual identification, but requiring an a priori of the potential species involved in the attack. Both approaches were validated by investigating two shark attacks that occurred in Reunion Island (southwestern Indian Ocean). According to both methods, each incident was attributed to a bull shark (Carcharhinus leucas), in agreement with suggestions derived from bite wound characteristics. Both approaches appear thus suitable for the reliable identification of species involved in shark attacks on humans. Moreover, microsatellite genotyping reveals, in the studied cases, that two distinct individuals were responsible of the bites. Applying these genetic identification methods will resolve ambiguities on shark species involved in attacks and allow the collection of individual data to better understand and mitigate shark risk.