Aversive learning was applied to affect the phototactic behaviour of the marbled crayfish. Animals initially showed negative phototaxis to white light and positive taxis to blue light. Using an aversive learning paradigm, we investigated the plasticity of innate behaviour following operant conditioning. The initial rate of choosing a blue-lit exit was analysed by a dual choice experiment between blue-lit and white-lit exits in pre-test conditions. During training, electrical shocks were applied to the animals when they oriented to the blue-lit exit. Memory tests were given to analyse the orientation rate to the blue-lit exit in trials 1 h and 24 h after training and these rates were compared with the pre-test. In general, animals avoided the blue-lit exit in the memory tests. When training was done three times, the long-term memory was retained for at least 48 h, although a single bout of training was also enough to form a long-term memory. Cooling animals at 4 °C or injection of cycloheximide immediately after training altered the formation of long-term memory, but had no effect on short-term memory formation. Administration of the adenylate cyclase inhibitor SQ22536, the PKA inhibitor H89, or the CREB inhibitor KG-501 immediately after training also blocked the formation of long-term memory, but had no effect on short-term memory formation. Thus, our pharmacological behavioural analyses showed that new protein synthesis was necessary to form long-term memories and that the cAMP/PKA/CREB pathway is the main signal cascade for long-term memory formation in the marbled crayfish.