The reflex response to stretch in most contracting human muscles includes both a short-latency, probably monosynaptic, excitatory component, and a longer-latency, polysynaptic excitation. However, it has been claimed that stretch of the jaw-closing muscles evokes only the short-latency response in masseter. This question was re-examined, using controlled stretches of varied rates and durations. Very brief, rapid stretches analogous to the stimuli used to investigate the 'jaw-jerk' reflex in earlier studies evoked a prominent excitatory peak in the electromyogram at monosynaptic latency excitation, but little or no longer-latency excitation. This response could be produced even by stimuli that were barely detectable by the subject. However, this prominent electrical response did not produce a measurable increase in biting force. In contrast, slower stretches evoked both a short- and a longer-latency excitatory response in the surface electromyogram, as in most limb muscles. It is shown that the absence of a long-latency excitatory response in earlier studies can be explained by the powerful reflex disfacilitation of the motoneurones that occurred at the end of the brief stretches used. Depending on the duration of the stretch, this disfacilitation is often sufficient to mask or abolish the long-latency reflex. The reflex response to stretches was not markedly affected by blocking the activation of mechanoreceptors around the teeth with local anaesthetic, indicating that receptors around the teeth cannot be playing more than a minor role in the response. The stretch-induced increase in force became greater as the velocity of the stretch decreased.