Drinking and thirst-motivated behaviour have traditionally been explained in terms of the rather simple concept of homeostasis. A homeostatic mechanism readily accounts for responses to acute changes in body-fluid levels. However, there are other factors regulating intake, for example, cues associated with eating, which interact with the time elapsed since last drinking and the availability of water. Future dehydration is avoided by behavioural hysteresis; a sudden reduction in fluid needs is not matched by an equivalent reduction in fluid intake. Another factor not explicable by traditional homeostasis is that, in general, drinking cannot be suppressed by water infusion. Nor are there rigid target values for body-fluid levels independent of the cost of obtaining water; when water is hard to get, a relatively low body fluid level is maintained, thus minimizing loss. On the basis of the results conflicting with traditional homeostatic theory, this paper tentatively suggests certain modifications toward a more realistic model of fluid ingestion.