Since the first recordings made of evoked action potentials it has become apparent that the responses of individual neurons to ongoing physiologically relevant input, are highly variable. This variability is manifested in non-stationary behavior of practically every observable neuronal response feature. Here we introduce the Neuronal Response Clamp, a closed-loop technique enabling full control over two important single neuron activity variables: response probability and stimulus-spike latency. The technique is applicable over extended durations (up to several hours), and is effective even on the background of ongoing neuronal network activity. The Response Clamp technique is a powerful tool, extending the voltage-clamp and dynamic-clamp approaches to the neuron's functional level, namely - its spiking behavior.