Meal patterns of rats (Rattus norvegicus) in a habitat with variable foraging costs were examined in a laboratory paradigm where individuals could initiate meals at any time by completing a predetermined number of bar presses (the procurement price) and then could eat any amount. In one phase the procurement price was constant from meal to meal; in another phase the procurement price varied from meal to meal randomly among a geometric series of five possible prices. The middle price of each variable-price schedule was equal to the price in one of the fixed-price schedules. As the fixed price, or the mean of the variable prices, increased across schedules, meal frequency decreased and meal size increased; daily intake was constant. Within variable-price schedules there was no effect of the just-paid price on the size of the meal, except in the highest-price schedules, when meal frequency was low and somewhat larger meals followed higher prices. Meals were no more variable in size within variable- than fixed-price schedules. The prandial correlations were not reliable, but the within-schedule variability in meal size was partially accounted for by diurnal variation, the effects of which were independent of procurement price. The results suggest that a global, rather than local, cost assessment, with a time window of about 24 h, influences the feeding patterns of rats. In the same way, food intake appears to be regulated over a global, rather than meal-to-meal, time frame, and thus, models that account for meal initiation in terms of the size of the prior meal or inter-meal interval, and meal size in terms of load, are inadequate to describe meal patterns in the freely feeding rat. © 1994 The Association for the Study of Animal Behaviour.