This paper studies the use of Unscented Kalman Filters (UKF) to estimate nonlinear dynamics and, specifically, adaptive determination of scaling parameters in these filters. Due to lack of analytic solution and use of numerical methods instead, the computational load of these filters increases drastically. In this paper, a new method is proposed based on Interactive Multiple Models (IMM) which has lower computational burden than previous methods. The performance of the proposed filter is evaluated by simulating a tracking scenario. Its accuracy, computational load and stability are compared with the numerical adaptive UKF filter and constant parameter UKF filter. The simulation results show that the proposed filter performs better than the constant parameter UKF and very close to the numerical adaptive UKF.