Context. It is difficult to measure feeding rates for most wild nocturnal mammalian herbivores. Thus, although koalas are a popular species to study, we have a poor understanding of their activity patterns and feeding ecology. Researchers often assume that the trees that koalas occupy during the day indicate feeding preferences, but they may better reflect preferred resting sites.

Aims. We recorded the activities of koalas, with an emphasis on their feeding, particularly when they fed, the trees that they fed from, the number of meals they consumed and the variability in these measurements.

Methods. We continuously monitored eight koalas by audio- and radio-telemetry for 14 consecutive 24-h periods each. We followed two koalas at a time and recorded the trees they visited, when, where and how long they fed, and the size and nutritional composition of the trees in the landscape.

Key results. Individual koalas varied in how many trees they visited, how many meals they ate and how long they spent feeding during each 24-h period. They preferred Eucalyptus globulus trees during the day, but fed mainly at night, with a preference for E. viminalis. The trees that koalas visited during the day were larger than those that they visited at night.

Conclusions. The trees that koalas occupied during the day were poor indicators of their diet preferences, whereas the daily feeding activities of individual koalas varied widely.

Implications. Predicting a koala’s diet from the trees it occupies during the day is fraught with error. Although the trees that koalas rest in are important in the species ecology for reasons other than feeding, we should refrain from using them to predict an animal’s diet. Because feeding activity is difficult to measure, it is probably best done indirectly by analysing leaf-cuticle fragments or waxes in faeces. The substantial day-to-day variation in koala activities also indicates that behavioural and physiological studies of koalas require long monitoring periods – a week or longer.