Skid-steered, tracked, tele-operated robots are used to perform high-risk critical missions such as bomb disposal under conditions deemed too risky to send a human. Often the robots carry heavy payloads that raise their centers of mass, increasing the risk of tip-over. Since it is often not feasible to send a human to right a toppled robot, tip-over is equivalent to mission failure. Hence, an autonomous behavior to prevent robot tip-over is desired. In this research, a simplified model of mobile robot dynamics permits separation of pitch and roll stabilization. Adaptive control is used to stabilize the appropriate angle based on the normalized tip-over measure. Experimental validation of this control is successfully demonstrated on an iRobot Packbot and a Segway RMP 440.