Several aspects of plant-animal interaction were studied using terrestrial molluscs and their food plants. 1. Seedling monocultures of 7 species of plants with different acceptability (measured previously using leaf discs) were grazed by slugs (Agriolimax Caruanae) for controlled periods. (I) The risks of death due to slug grazing varied with plant size and growth form. Acceptability was a poor predictor of the risks of death due to grazing. (ii) Seedlings were killed by slugs, remained intact or were damaged but not killed; the acceptability to slugs of mature leaves was a poor predictor of the damage done to seedlings. (iii) Plants of at least some of the species had an increased risk of death after grazing. Those grazed plants that survived were usually smaller. (iv) It is suggested that for most species there is a 'critical size of escape' when the plants are able to tolerate grazing without being killed. 2. Monocultures of Capsella bursa-pastoris and Poa annua were grown, at densities high enough for plants to interfere with each other's growth and survival, in the presence and absence of slugs (A. caruanae). (1) Slugs behaved quite differently towards the two plant species. They ate whole, or parts of, leaves of C. bursa-pastoris, but rarely killed the plants. They ate P. annua (only after a 'starving threshold') by chewing through shoots at ground level, 'felling' but not consuming the shoots; many of the damaged plants of Poa died. Grazed and ungrozed populations - II - of both species conformed to the -3/2 thinning law. (ii) The compensatory effects of competition and predation are discussed. 3. Populations of Capsella bursa-pastoris were grown at two densities in a background of plant diversity given by the seedlings emerging naturally from soil collected from areas where the slugs Agriolimax caruanae and A. reticulatus are abundant. The populations were grazed by both slugs and the plants of Capsella were monitored for most of their life cycle. (i) C. bursa-pastoris was clearly the species most preferred by slugs. (ii) Grazing of Capsella by slugs affected plant survival, rosette diameter, number of flowering and fruiting plants and the number of capsules per plant. The effects were almost exclusive to the populations of low density. The reproductive potential of Capsella was also affected by defloration., (iii) It was suggested that the diversity of the vegetation that is relevant to a given herbivore depends on its search and range and that studies on the role of herbivores on a population must take into account the diversity of the whole community (inter- and intraspecific). 4. The intraspecific variation in plant acceptability was studied by examining the response of slugs to the cyanogenesis polymorphism in Trifolium repens. (i) Molluscs failed to distinguish between cyanogenic and ocyanogenic leaf discs or detached clover leaves but experiments with intact growing plants, as well as field surveys, showed clearly that molluscs preferred the acyanogenic morphs. The pattern of grazing was characterised by an excess of nibbling on the cyanogenic plants and an excess of heavy damage on the acyanogenic forms. - III - (ii) Animals fed on monotonous diets of lettuce grew considerably faster than slugs feeding on diets of cyanogenic and acyanogenic clover leaves but there was a marginally lower rate of growth of the slugs fed on cyanogenic leaves. (iii) There was a significant association between the spatial distribution of mollusc activity and clover morphs in the field: areas of intense mollusc activity had an excess of cyanogenic morphs while in areas of low mollusc activity the acyanogenic morphs were represented in excess. (iv) Homogenates mdde from the digestive tract of slugs produced cyanogenic reactions with clover leaves that had cyanoglucosides but lacked hydrolising enzymes. (v) A tentative coevolutionary model of the clover-mollusc interaction is suggested. 5. The performance of transplanted cuttings of cyanogenic and acyanogenic morphs was observed in the field for 1 year. (i) There was some superiority of the acyanogenic morphs for several growth parameters and plant survival during the growing season of the sward. (ii) There was much heterogeneity between clones of both morphs with respect to their susceptibility to grazing by several herbivores; nevertheless, there was a consistent preference for the acyanogenic forms by molluscs. Weevils and sheep showed preference for one morph at one time of the year, and for the other morph at another time. (iii) Damage by the rust Uromyces trifoii was almost en%rety confined to the cyanogenic morph. - Iv - (iv) It is concluded that studies on the interaction between clover and mollusc must also take into account the intraspecific (or even intraclonal) variation in reaction of clover to grazing, pathogenic infection and response to competing neighbours.