The Information Matrix Fusion (IMF) algorithm for nonlinear, asynchronous (with arbitrary local tracker sampling times for full rate as well as reduced-rate communication) and heterogeneous systems is presented. The heterogeneous estimates from local trackers are in different state spaces with different dimensions and are related by a non-linear and noninvertible transformation. The main application of these results is the fusion of tracks from radar and infrared/electrooptical sensors. Different from Track-to-Track Fusion, the IMF does not re-quire the cross-covariance between the local estimation errors. The performance of the proposed algorithm is shown via simulation based on Monte Carlo runs and is compared with the optimal solution—full-rate centralized fusion for both full-rate fusion and reduced-rate fusion for heterogeneous and asynchronous sensors.