A dead-reckoning scheme appropriate for skid-steered mobile robots is introduced to serve in a wider scheme for simultaneous localization and map-building. It is based on internal sensors (inertial data and odometry) only. The information from an experimentally derived kinematic model and an onboard inertial navigation system (INS), after a necessary pre-filtering stage, is fused using a simple and fast modified Kalman filter. We verify our approach with large scale experiments in outdoors structured environments and variable terrains, following paths with steep turns and variable velocity.