A variety of forms of painful stimulation were delivered to human subjects in order to determine whether therapeutic dosages of systemic morphine might produce significant attenuation of some forms of phasic pain that are tolerable for experimental usage. Consistent with previous reports, simple application of thermal or electrical energy to the skin (for 3 sec) produced sensations of pain that were not significantly reduced by prior administration of morphine. Similarly, subjects that were trained to focus their attention on the magnitude of the immediate (first) pain sensation evoked by brief electrical or mechanical stimulation did not report reduction by morphine of pain attributed to conduction in myelinated peripheral nociceptors. In contrast, the magnitude of late (second) pain sensations produced by brief pulses of electrical, thermal or mechanical stimuli to the same subjects was consistently reduced significantly by doses of 5 or 10 mg of morphine. The simplest interpretation of the effect on second pain intensity is that morphine preferentially attenuates input from unmyelinated nociceptors. This conclusion was reinforced by an experiment in which chemicals were applied to the skin. Morphine reduced pain produced by capsaicin (presumed to selectively excite unmyelinated peripheral afferents) but did not diminish pain elicited by bradykinin (presumed to excite A delta and C nociceptors). Comparing long duration pains from chemical stimulation (lasting in excess of 5 min) with briefer pains elicited by 50 msec to 3 sec of stimulation did not support the notion that morphine acts selectively on tonic pain. Also, after-sensations that could be discerned following second pain were not eliminated by morphine, and paired pulse facilitation of first pain sensations remained after administration of morphine, indicating that temporal summation is not preferentially reduced. Regardless of duration, frequency or latency, pain arising exclusively from unmyelinated nociceptors was attenuated substantially, but other elicited sensations were not reliably affected. For example, detection thresholds for warmth were unaffected by morphine, demonstrating that input from all unmyelinated afferents is not reduced.