In cats decerebrated at the premammillary level, the incremental stiffness of actively contracting soleus muscle was measured during ongoing changes in force throughout the physiological length and tension range. Rectangular incremental length perturbations 1 mm in amplitude and 200-500 ms in duration were applied once per second by a torque motor, which otherwise held the muscle isometric. The resulting force increment was divided by the applied length increment to obtain the incremental stiffness. Two separate mechanisms contribute to the incremental stiffness: the intrinsic stiffness of actively contracting fibers undergoing stretch and the recruitment of additional tension via the stretch reflex. The relative contribution of each mechanism was determined by comparing the properties of muscle under reflex regulation with the response of areflexic muscle in the anesthetized preparation. In the latter case a range of operating forces was generated by electrical stimulation of the muscle nerve. In areflexic muscle the short-range intrinsic stiffness, attributed to elastic deformation of cross bridges, causes an early rise in force concurrent with the onset of stretch. A decline in force, attributed to detachment of cross bridges, follows when the amplitude of stretch exceeds their elastic limit. Subsequently a new force level is reached and maintained, reflecting reformation of cross-bridge bonds. The amplitudes of the early peak and of the final force increment 200-500 ms later are proportional to the operating force. With the stretch reflex present, the stiffness of the muscle is increased for all operating lengths and forces within the physiological range. The additional tension recruited by the reflex varies nonlinearly with operating force, peaking in the middle of the range and declining toward zero for high forces. Reflex effects combine with intrinsic muscle properties to produce total stiffness values that are uniform within .+-. 15% for operating forces 25% of maximum or greater. With reflex present, stiffness varies little with operating length. For matched operating conditions the magnitude of the incremental stiffness is remarkably invariant. No spontaneous changes in the gain of the stretch reflex were observed in these preparations. Whereas force increments caused by stretch are quite reproducible at matched operating points, the amplitude of the early synchronous (presumably monosynaptic) EMG [electromyogram] peak is highly variable and not correlated to force. Average EMG increments and reflexly mediated force increments caused by stretch are linearly correlated for low operating forces but diverge for high forces, suggesting saturation in the contractile apparatus rather than in the segmental reflex. Transient decreases in reflex gain not observed under other circumstances can be produced by electrical stimulation of selected ipsilateral and contralateral peripheral nerves and brain stem structures with trains of brief, low-intensity stimuli. After about 10 s incremental stiffness values return to normal levels in spite of continued stimulation. The role of the stretch reflex in the regulation of muscle stiffness is evaluated. Implications regarding the regulation of joint stiffness during locomotion and during execution of voluntary motor tasks are discussed.