Feeding behaviour has been studied in rats (Rattus norvegicus Berkenhout) kept in small cages in which a choice of foods was offered. Most of the rats were wild, but some albinos were used for comparison. The main intention of the experiments was to elucidate the component activities which make up feeding behaviour; but some observations on food preferences were made. When offered minced liver, wheat grains, flour and sugar, rats usually ate some of each food; sugar was eaten least, and sometimes little of e i t h e r flour or wheat was taken. When offered the alternative of wheat mixed either with cod liver oil or aniseed oil, as against wheat mixed with arachis oil, consumption was almost confined to the arachis oil mixture. This was true even when, as nestlings, the rats had been given only the distasteful mixture to eat. Both wild and albino rats show very marked exploratory and sampling behaviour during feeding. Evidence is presented that this is a pattern in its own right, and not merely appetitive. Its function is to make possible the rapid learning of topography, and especially the whereabouts of food, water and shelter. Sampling is also important in enabling rats to associate particular physiological actions, whether nutrient or toxic, with different foods. Exploration and sampling occur even when the situation and the foods are familiar, and regardless of age or sex. Continued sampling makes possible regular reinforcement of learned preferences. It also enables rats to change rapidly from a familiar but distasteful food to an unfamiliar but palatable one. Sampling is not stopped by 'satiation', but is released by it. In wild rats, but not albinos, there is highly developed flight and avoidance behaviour. This tends to reduce the dangers of exploratory behaviour. Both wild and tame rats tend to feed in cover when this is practicable. This has indirect social effects, since one rat may rob another in the nest. Social influences on feeding are only incidental. Often they are a product of associative learning. There is no parental guidance of young, nor imitation by young of adults. Displacement grooming was seen when hungry rats were deterred from feeding by the unfamiliarity of the situation; and in other frustrating circumstances. It is concluded that much of the feeding behaviour of wild rats can be interpreted as a resultant of the opposed effects of exploration and sampling on the one hand, and flight and avoidance on the other.