Commissural interneurons in rhythm generation and intersegmental coupling in the lamprey spinal cord. To test the necessity of spinal commissural interneurons in the generation of the swim rhythm in lamprey, longitudinal midline cuts of the isolated spinal cord preparation were made. Fictive swimming was then induced by bath perfusion with an excitatory amino acid while recording ventral root activity. When the spinal cord preparation was cut completely along the midline into two lateral hemicords, the rhythmic activity of fictive swimming was lost, usually replaced with continuous ventral root spiking. The loss of the fictive swim rhythm was not due to nonspecific damage produced by the cut because rhythmic activity was present in split regions of spinal cord when the split region was still attached to intact cord. The quality of this persistent rhythmic activity, quantified with an autocorrelation method, declined with the distance of the split spinal segment from the remaining intact spinal cord. The deterioration of the rhythm was characterized by a lengthening of burst durations and a shortening of the interburst silent phases. This pattern of deterioration suggests a loss of rhythmic inhibitory inputs. The same pattern of rhythm deterioration was seen in preparations with the rostral end of the spinal cord cut compared with those with the caudal end cut. The results of this study indicate that commissural interneurons are necessary for the generation of the swimming rhythm in the lamprey spinal cord, and the characteristic loss of the silent interburst phases of the swimming rhythm is consistent with a loss of inhibitory commissural interneurons. The results also suggest that both descending and ascending commissural interneurons are important in the generation of the swimming rhythm. The swim rhythm that persists in the split cord while still attached to an intact portion of spinal cord is thus imposed by interneurons projecting from the intact region of cord into the split region. These projections are functionally short because rhythmic activity was lost within approximately five spinal segments from the intact region of spinal cord.