This report describes various aspects of classification and datafusion with an eye to principal methods. A discussion on generic methods is followed by a short description of commonly used classification methods. Datafusion is discussed in terms of feature and decision fusion. Three FOI projects are studied in more detail. The scenarios are composed of simulated landscapes with inserted models of vehicles, or show real combat vehicles moving in natural terrain. The sensors may be either simulated or real. In the project Multi-Target-Seeker target attributes have been extracted from segmented IR scenes and high-resolution range profiles from a radar. The results show, that using an artificial neural net, targets are classified correctly more often with combined IR and radar data, given their uncertainty. In the project Interactive Ground Sensor Network the tracker fuses kinematic data with ID information on acoustic sensorsdata, making association of target-observation easier in a complex scenario. In the project IR/mm-wave radar the number of possible targets can be strongly reduced by means of an IR contrast filter, scaled by range from a radar.