Joint target tracking and classification is a challenging problem where the class of a target must be estimated in addition to its kinematic states, such as position and velocity. This problem is of special importance both in civilian and in military domain, where target classification plays an important role in the decisions that an operator makes. Moreover, when several sensing options are available for performing joint target tracking and classification then a sensor management problem arises in addition to the joint tracking and classification problem. For addressing this sensor management problem, we propose managing the uncertainty in the threat-level of a target under observation. Since threat is a context-sensitive quantity, it can be defined in different operational contexts both civilian and military. This makes threat-based sensor management for joint classification and tracking a promising alternative to standard sensor management schemes that can be found in the literature. In order to support the latter statement and demonstrate the potential of our idea, we show simulated examples from both domains.