In this work, we present a strategy for intercepting a highly maneuverable evader while exploiting delayed and imperfect state information. The imperfect sensing arises from uncertain sensor measurement, data losses, communication delays and conflict resolution among objectives such as communication and detection. This problem is further complicated by the challenge of predicting evader's plan-of-action with incomplete and delayed information about evader's current states. To solve this problem, we suggest a probabilistic approach using a) probability distribution of evader's feasible trajectory envelope that can be reached in sequential time-steps and, b) evader's possible motivations such as intelligent evasion and heading to valuable landmarks and way outs. The method is not critically based on exact model of the evader's behaviors and robust to imperfect measurement about evader's location. High uncertainty in detection and estimation of evader's location is exploited by probabilistically modeling its motion and action plan. We have demonstrated performance of this approach in simulations using trade off between success of interception and rate of uncertainty in sensing and communication lags.