1. The physiology of large monopolar cells (LMCs) in the lamina of the visual system of the droneflyEristalis was investigated by recording the intracellular responses to light flashes, in the presence or absence of intracellularly injected current. 2. In the dark, the steady-state voltage-current relationship is approximately linear for small excursions of membrane potential (Fig. 1). In different recordings LMC dark input resistance ranged from 20 to 45 MΩ. 3. As in other fly species, the intracellular response to a flash of light from a point source consists of a hyperpolarizing transient at ‘light on’, followed by a small, sustained plateau (usually hyperpolarizing), and a depolarizing transient at ‘light off’ (Fig. 2). 4. In some LMC recordings the injection of steady current affects primarily the ‘on’ response, and not the ‘off’ response. The amplitude of the ‘on’ transient is larger in the presence of depolarizing current and smaller in the presence of hyperpolarizing current (Figs. 2, 3), implying a transient decrease of input resistance during the ‘on’ response. Recordings of this nature tend to be associated with comparatively low values of dark input resistance (Fig. 5). In other recordings the injection of steady current affects primarily the ‘off’ response, and not the ‘on’ response. The amplitude of the ‘off’ transient is larger in the presence of hyperpolarizing current, and smaller in the presence of depolarizing current (Fig. 4), implying a transient decrease of input resistance during the ‘off’ response. Recordings of this nature tend to be associated with comparatively high values of dark input resistance (Fig. 5). 5. In comparison with the ‘off’ response, the ‘on’ response is much more labile (Fig. 6). This observation, taken together with the above data, suggests that the ‘on’ and ‘off’ responses of the LMC are generated at independent, spatially-segregated sites. 6. We propose a model of LMC function in which the sites for ‘on’ and ‘off’ response generation are spatially separated. This model accounts for the variability in effectiveness of extrinsic current injection on the ‘on’ and ‘off’ responses. We suggest that in distal penetrations (i.e. near the photoreceptor synapses) injected current affects primarily the ‘on’ response, whereas in proximal (i.e. axonal) penetrations it is the ‘off’ response that is most affected. The range of experimentally determined values of dark input resistance, and their relation to the effect of extrinsic current on the ‘on’ or ‘off’ response (Fig. 5), agrees with the predictions of the model that dark input resistance varies with electrode position along the LMC. Thus, we suggest that low values of input resistance (ca. 20 MΩ) correspond to distal penetrations, and high values (ca. 45 MΩ) correspond to proximal penetrations. 7. In 3 cells, the injection of steady hyperpolariting current led to the occurrence of spontaneous depolarizing events and action potentials in the dark (Figs. 7, 8). In 2 of these cells light increased the frequency of events and action potentials; in another it had the opposite effect. We suggest that the light-evoked response of the LMC under these conditions reflects the combined effect of a hyperpolarizing input at the receptor synapse, and at least one additional, more proximal input driving the depolarizing events and spikes. This is the first report of impulse activity in LMCs, which have hitherto been assumed to encode visual signals in a purely analog fashion.