1. The identification of numerous functional connections among medullary respiratory related neurons led us to postulate specific short time scale correlations among neuronal spike trains consistent with 1) inhibition of rostrolateral medullary augmenting expiratory (E-AUG) neurons by other E-AUG cells, and 2) inhibitory control of the postulated E-AUG neural network by decrementing expiratory (E-DEC) neurons. Recent observations of reduced rostrolateral E-AUG cell activity following stimulation of sensory afferents raised the additional question of whether reflex control mechanisms use the identified functional connections to mediate their effects. 2. Experiments were conducted on 42 anesthetized, paralyzed, bilaterally vagotomized, artificially ventilated cats. Impulse trains of two or more respiratory related neurons of the lateral medulla, including at least one rostrolateral medullary E-AUG neuron, were simultaneously recorded together with phrenic nerve efferent activity. Most neurons were tested for spinal axonal projections with antidromic stimulation methods. In some experiments, the central cut end of the right vagus nerve was electrically stimulated during the expiratory interval. Data were analyzed using cycle-triggered histograms, auto- and cross-correlograms, logical cross-correlograms, snowflake scatter diagrams, and peristimulus time histograms. 3. Seven of 73 pairs (9.6%) of ipsilateral E-AUG neurons exhibited short time scale correlations in firing probability. Two rostral pairs had coincident increases in activity. Five pairs were characterized by a reduction in activity in a rostral neuron following spikes in the other cell; concurrent serial inhibition among one set of three E-AUG neurons was indicated. These five pairs, together with three other similarly correlated pairs described previously, were studied further: spikes recorded during the first and second halves of the expiratory (E) phase were analyzed separately. This phase segmentation unmasked multiple correlations implying reciprocal inhibition between one pair of E-AUG neurons. Short time scale troughs in correlograms generated from late E-phase spike trains had significantly (P less than 0.05) greater detectability indices than troughs in early E-phase correlograms. Multineuron correlations revealed two cases in which the firing probability of the target E-AUG neuron was reduced less when only reference E-AUG cell spikes coincident with spikes in a third, E-DEC, neuron were used as trigger events. 4. Enhanced rostral E-AUG neuron firing probabilities coincident with or following impulses in ipsilateral E-DEC neurons were detected in 6 of 94 (6.4%) rostral pairs.(ABSTRACT TRUNCATED AT 400 WORDS)