The existence of personality in nonhuman animals is widely accepted, but the development of valid and reliable methods to measure and interpret it is ongoing. Personality can be important in selecting horses for particular roles and has the potential to improve welfare and safety for riders and horses. This is especially important for challenging roles such as police work. A custom-made behavioral test protocol was trialed with the Belgian mounted police, aimed at complementing the existing subjective rating of potential new police horses by experienced mounted police officers. A sample of 48 active police horses was used to assess the feasibility of the protocol and to establish an initial reference population for future comparison. The protocol was a continuous program of 9 tests in an indoor arena, including tests of responses to solitary activity and sudden stimuli and interactions with humans and conspecifics. This program yielded 10 Likert scores and 115 continuous variables, including locomotory responses, rolling, whinnying, and distance categories where relevant and feasible. A first test for the discriminative power of the behavioral selection protocol was the comparison of the well-suited to the less-suited police horses in the sample. Only 6 of the 125 variables, divided over 4 of the 9 tests, showed a significant difference. Overall, this study showed that a custom-made behavioral testing protocol is feasible for a working mounted-police unit. Although it established an initial reference population, the protocol did not allow a clear distinction between the well-suited and less-suited police horses currently active. As such, this study represents the first phase of the development of a valid and reliable protocol to support selection of police horses with behavioral test results.