Simulations are presented of Automatic Target Recognition (ATR) of ground targets with High Range Resolution Radar (HRRR), operating at 35 GHz. In the considered application the radar is a possible choice in the multisensor package of a missile seeker, together with an IR sensor or a laser. Only the ATR properties of the radar channel, seen in isolation, are considered in this study. Down-range radar cross-section profiles are used as basis for the classification of wheeled or tracked military targets of six possible types. Homogeneous ground clutter (grass) is ssumed. ome methods are studied for reducing the amount of data involved in the classification process. These consist f the use of central moments of range profiles, and of two integration schemes. The latter include simple profile averaging, or generating characteristic profiles from a Singular Value Decomposition of the matrix of collected data. Accumulated classification of such integrated data is assumed with a view to improving recognition performance with the missile approach. A moderate decline of the classification occurs when central moment data are used, in the weak clutter case studied. There is no significant difference between the two integration schemes. The accumulated classification displays an expected successive improvement during the flight. Central moments of range profiles produced with two superresolution algorithms (MUSIC and Capon´s method) do not improve the classification. The use of MUSIC delivers consistenly the weakest performance, since amplitude information is lost.