In this paper two nonlinear estimation techniques, i.e. particle filter (PF) and unscented Kalman filter (UKF), are used for estimation of the states of a passive homming missile. The estimated states including the range between missile and target are used as the elements of proportional navigation guidance (PNG) law. Previous methods such as extended Kalman filter (EKF), measurements based on line of sight angle and its rate of change have demonstrated some difficulties due to limitations of extended Kalman filter as well as unobservability of the states of the system. Besides, in the paper the problem of observability of states of the system is investigated and unobservability of some of the states of this system is resolved by proposing a new measurable variable which can be used in real implementations of such missiles. The performances of these two nonlinear estimation techniques are compared with that of extended Kalman filter from root mean square error point of view by Monte-Carlo simulations. The computational complexity and the question that the algorithms are realizable or not are also considered in this study.