The limits on maximum information that can be transferred by single neurons help us to understand how sensory and other information is being processed in the brain. In this paper we approximately calculate the information capacity of the perfect integrate-and-fire neuronal model in dependence on the stimulus range, assuming the simplest form of temporal coding scheme. We couple the information transfer with metabolic cost of neuronal activity and we find that the optimal information per metabolic cost ratios may occur for a relatively small stimulus range.