The neural responses of 456 single units were recorded in parietal cortex of behaving monkeys during a haptic delayed matching-to-sample task. (1) In areas 2 and 5 together, 22% of the neurons were activated by the auditory cue that signalled the beginning of a trial. Virtually all of these cells were also activated during the arm movements required by the task. These neurons, showing both auditory-related and movement-related responses, may function in sensorimotor integration. (2) Responses related to arm projection frequently began before movement onset, sometimes as much as 320 ms before. Such "premovement" responses were approximately equally common, and showed the same latency distribution, in areas 2, 5a, and 5b. (3) There was a topographic rostral-to-caudal gradient of decreasing neural responsiveness to the animal's manipulation of the cue (sample) objects. Eleven percent of manipulation-activated cells responded preferentially to one of the sample objects. (4) Many cells showed sustained (greater than 3 s) activation during the delay period (the time between handling of the sample object and palpation of the choice objects), even though at that time the monkey was sitting quietly and without stimulation. (5) Cells with sustained activation throughout most or all of the 18-s delay period were rare in all areas tested except area 5a. These cells, especially those that were preferentially activated depending on which sample object was palpated, may function in the temporary retention of haptic attributes. (6) The population of cells activated during sample manipulation was largely distinct from the population of cells showing sustained activation during the delay period. These two cell populations may represent different but complementary aspects of haptic perception. (7) The most common response during the delay period was sustained inhibition. This may be an expression of a nonspecific mechanism for decreasing background noise and enhancing neural responses to an anticipated perceptual event. (8) Relatively little evidence was found to support a functional distinction between the neural response properties of areas 2 and 5a. This suggests that area 2 may be at a higher level in the somatosensory hierarchy of the posterior parietal cortex than usually considered.