1. Cross-correlation analysis of simultaneously recorded spike trains can be used to gain insight into functional interactions among neurons. In this paper, we report on cross-correlation analysis of neuron pairs in the dorsal cochlear nucleus (DCN) of the cat. Neuron pairs were isolated with two independent electrodes, which allow systematic study of the effects on correlation of distances between units and differences in their best frequencies (BFs). The data in this paper were obtained from 51 pairs consisting of two neurons of the same type. 2. Cross-correlograms were obtained for 35 pairs composed of type IV units, which are recorded from the principal cells of the DCN. Pairs of type IV units with correlated activities give cross-correlograms with increased correlation near zero delay. This feature is called a central mound (CM) and most likely results from shared excitatory or shared inhibitory inputs. 3. Records of spontaneous activity were obtained from 31 pairs of type IV units. Six of these pairs have correlated spontaneous activities. All six pairs have BFs that differ by less than 0.2 octaves. The shared input inducing these correlations must be a spontaneously active and tonotopically organized projection, like the auditory nerve. Type II units, thought to be DCN inhibitory interneurons that project to type IV units, are not spontaneously active, and thus cannot be the cause of correlated spontaneous activity. Similarly, cochlear granule cells, whose axons project orthogonally to the tonotopic sheets of DCN, cannot be the cause of correlated spontaneous activity because their projection is not confined tonotopically. 4. Stimulus-driven activities were studied for 12 type IV pairs that have uncorrelated spontaneous activities. Five of these pairs have correlated driven activities, with CMs whose sizes depend on the frequency and sound level of the acoustic stimulus. A frequency vs. sound level correlation response map shows the V-shaped tuning properties of the correlation-inducing mechanism. The properties of stimulus-driven correlation in these type IV pairs are consistent with the hypothesis that the correlation is induced by shared input from DCN type II units, although this is not the only possibility. 5. All six type IV pairs with correlated spontaneous activities have correlated driven activities. In five of these pairs, the degree of correlation decreases from its value with spontaneous activity when a low-level acoustic stimulus is applied. Three of these five pairs were tested at higher stimulus levels.(ABSTRACT TRUNCATED AT 400 WORDS)