As ballistic targets reenter the earth’s atmosphere, they undergo rapid deceleration and may perform spiraling motion that can make accurate tracking very difficult. A variety of filter structures have been proposed for this problem, including variants of the Kalman filter. The present work employs a new target dynamic model combined with an unscented Kalman filter to yield an effective tracking solution. The new target model includes, in addition to position and velocity states, a modified drag coefficient, the target spiral frequency, and two harmonic states describing the periodic rotation of the target body around its velocity vector. The spiral frequency was allowed to have a general variation with time. The unscented Kalman filter with new target model produced very satisfactory tracking for targets with varying frequency. The sensitivity to angle measurement accuracy using a reduced set of measurements was evaluated and a threshold value was determined. Finally, tracking performance was confirmed by Monte Carlo analysis.