This paper attempts to determine the effect on the number of prey eaten by predators of the addition of the component "avoidance learning by prey" to a computer model of the predation process developed by Holling. Generality was retained by concentrating upon a basic aspect of the prey's behaviour, its distance of reaction to an approaching predator. The zebra danio (Brachydanio rerio), a small freshwater fish, was used as an analogue of a general vertebrate prey. The predator used was the largemouth bass (Micropterus salmoides).Previous work (Dill, 1973b) showed that prey reactive distance increased with increasing experience with the predator. In the present study, this increased prey reactive distance is shown to increase predator pursuit time and hypothesized to decrease predator pursuit success. These relationships were expressed mathematically and built into Holling's (1965, 1966) model of the predation process, along with an equation describing the way in which reactive distance increases following an unsuccessful attack. Other changes necessitated in the model by the addition of the avoidance learning component included: a) Modifications of the calculation of search time to remove a previously implicit time spent unsuccessfully pursuing prey, and to correct the density of prey to account for those whose reactive distances exceed that of the predator and are therefore not susceptible to discovery; b) Addition of a new subroutine (CHASE) to calculate pursuit time, unsuccessful pursuit time, pursuit success, and strike success; c) Changes in subroutine ADCOM to assign prey to different classes (with different reactive distances) according to the number of times they have been unsuccessfully attacked; and d) Addition of a stochastic element via random numbers to determine the class to which an attacked prey belongs, the time to refuge, and the predator's strike success.Simulation was used to explore the consequences of these additions. The capability of learning substantially increased the prey's probability of surviving subsequent attack. Addition of an avoidance learning component caused declines in the predator's functional responses to both prey and predator density. The new component was also suggested to decrease the predator's numerical response to prey density and to increase the probability of stability in a predator-prey interaction.