We re-examined the issue of how a subject's intention to react to a joint perturbation may modulate the long-latency M2 stretch reflex response. The experiments were done on the flexor pollicis longus muscle (FPL) of the human thumb, for which there is evidence that its M2 reflex response is mediated, at least in part, by a pathway that traverses the motor cortex. The participation of the cerebral cortex in the genesis of the M2 reflex response may allow for a modulation of its amplitude, based on the intention of the subject. To test whether the M2 response is genuinely modulated by the subject's intention, we examined the magnitude of this response as a function of the FPL background level of activation, measured by the surface rectified and filtered EMG. The subject was instructed either to oppose the perturbation as quickly as possible, not to react, or to relax as quickly as possible after the onset of the perturbation. The time integral of the long latency FPL EMG response, computed between 40 and 70 ms following the onset of stretch, was plotted against the mean torque produced by the distal inter-phalangeal joint of the thumb, or against the mean background FPL EMG. There were no significant differences in the FPL M2 EMG responses for different instructions. The amplitude of the reflex response was dependent only--in an approximately linear manner--on the background level of muscle activation. The total joint stiffness (intrinsic plus reflex) was also calculated for each combination of instruction and background torque.(ABSTRACT TRUNCATED AT 250 WORDS)