A complete understanding of the cellular mechanisms underlying the formation of associations between stimuli, as occurs during classical conditioning, requires an understanding of the non-associative effects of the individual stimuli. The siphon withdrawal reflex of Aplysia exhibits both non-associative and associative learning when a tactile stimulus to the siphon serves as a conditioned stimulus, and tail shock serves as an unconditioned stimulus. In this chapter we describe experiments which examine the non-associative effects of tail shock at three different levels of analysis. At a behavioural level we found that the magnitude, and even the sign of reflex modulation induced by tail shock depended critically on three parameters: (i) the state of the reflex (habituated or non-habituated); (ii) the strength of the tail shock, and (iii) the time of testing after tail shock. Specifically, when non-habituated responses produced by water jet stimuli to the siphon were examined, tail shock produced transient inhibition 90 s later; facilitation of non-habituated responses (sensitization) only emerged after a considerable delay of 20-30 min. When habituated responses were examined, tail shock produced immediate facilitation (dishabituation); the amount of facilitation was inversely related to the strength of tail shock, with stronger shock producing no dishabituation. At a cellular level it was found that the complex excitatory postsynaptic potential (EPSP) in siphon motor neurons produced by water jet stimuli to the siphon provides a reliable cellular correlate of several of the non-associative effects of tail shock that we observe behaviourally. When non-decremented complex EPSPS were examined, strong tail shock produced transient inhibition at a test 90 s after shock. When decremented complex EPSPS were examined, weak tail shock produced immediate facilitation whereas strong shock produced no facilitation. Moreover, in these experiments tail shock had differential effects on the complex and monosynaptic inputs to siphon motor neurons, suggesting that in addition to the well-studied monosynpatic input, other elements in the neural circuit for siphon withdrawal may contribute to the modulation induced by tail shock. At a pharmacological level we found that the neuromodulator serotonin could reliably mimic some of the effects of tail shock. Specifically, brief application of serotonin produced transient inhibition of both the siphon withdrawal reflex and of nerve shock elicited complex EPSPS in siphon motor neurons. Interestingly, serotonin simultaneously produced facilitation of the monosynaptic connection from sensory to motor neurons. This dissociation in the effects of serotonin on complex and monosynaptic EPSPS suggests that serotonin may act at multiple synaptic loci to produce the net inhibition in complex synaptic input. Taken collectively, these results suggest that the diverse behavioural effects of tail shock may be mediated by modulation at multiple sites in the neural circuit for siphon withdrawal. Understanding the cellular mechanisms that underlie these diverse non-associative effects of tail shock will be important in formulating comprehensive cellular models of associative learning in this reflex system.