A relatively shorttime ago the individual neuron was viewed as functioning by means of the simple summation of brief inhibitory or excitatory events. The complexity of the nervous system was the outcome largely of the connections between neurons. Recent studies have uncovered a new set of phenomena that indicate that relatively complex information processing may occur at the level of the individual neuron. For example, rather than producing additive effects, synaptic inputs can produce multiplicative effects that serve to alter the gain of the system. In addition, synaptic inputs may be able to alter specific classes of other inputs selectively. One could envision that each class of synaptic input to a cell could be selectively depressed or enhanced by a corresponding modulatory input. Since the modulatory actions can be transmitted intracellularly via second messengers, an extensive array of presynaptic connections may be unnecessary. It remains to be determined which of the modulatory phenomena currently reviewed are functionally important and which are only pharmacological or experimental curiosities. Are there any common attributes of the various synaptic phenomena that have been termed modulatory? The most common features of modulatory synaptic effects are long-duration of action and contingent action. Contingent action refers to the property that modulatory transmitters often have little or no effects in themselves, but instead they alter the effects of other events. Long-duration and contingent action endow modulatory effects with properties ideally suited to the control of behavioral modulations such as learning, motivational state, arousal, and sensitization. While there is no necessary connection between behavioral modulation and neural modulation, the available evidence from invertebrates suggests that there often is a connection. It is likely that examples of neuromodulation in vertebrates also will be tied to behavioral modulation.