Abstract The conservation of hollow-dependent fauna is a major forest management problem in natural eucalypt forests managed for wood production. Eucalypt forests support a diverse fauna that utilise hollows in trees. Hollow-bearing trees are a component of forest structure often lost or substantially modified as a result of timber harvesting operations, especially those that are intensive (like clearfelling) and take place on a short rotation (< 120 years) One approach adopted to conserve populations of hollow-dependent fauna in eucalypt forests has been to retain suitable trees in logged areas. This paper reviews a wide range of issues associated with the retention of such trees ranging from: (1) the characteristics of trees which should be selected for retention; (2) the number and spatial arrangement of retained stems; (3) the need for recruitment of new hollow-bearing trees to ensure there is a perpetual supply of such resources; and (4) how best to protect those trees that are retained. We also consider factors such as the inter-relationships between tree retention strategies and silvicultural requirements, and on-site tree retention strategies and forest management at the landscape-scale. A major conclusion from our study is that the array of issues associated with the retention of trees with hollows in eucalypt forests is considerably more complex than reflected by current management practice. Indeed, it appears that most existing prescriptions for the retention of trees with hollows in logged sites may not ensure a perpetual supply of a range of types of hollow trees or provide the necessary habitat conditions to support viable populations of some species of hollow-dependent fauna. This is because current prescriptions typically do not take account of factors such as: the prolonged periods until hollows first begin to develop in Eucalyptus trees; mortality and collapse amongst retained stems; and the need to supply trees with suitable characteristics and in a suitable configuration to meet the requirements of the full range of vertebrate taxa that utilise this resource. Taking these factors into account, we conclude that only modified partial cutting systems would adequately provide for the needs of all cavity-using species across logged sites in eucalypt forests. A major problem confounding attempts to develop more informed prescriptions for the retention of trees in logged areas is a lack of data on key topics such as stem longevity and hollow ontogeny, mortality of retained trees on logged sites and cavity use by animals. There is scope for some of these data to be readily gathered in routine timber assessment and inventory, stand regeneration surveys and fauna and vegetation surveys. We recommend that forest and wildlife management agencies review their field survey methods and include some new measures among the range of data collected. There is also considerable merit in instigating new tree and forest modelling projects to simulate long-term stand conditions in wood production eucalypt forests. Such studies would guide forest and wildlife managers in the development of tree retention strategies needed to create and maintain stand characteristics suitable to support populations of hollow-dependent wildlife over not one, but many rotations. Given the uncertainty associated with the effects of these operations on hollow-dependent fauna, and continued timber harvesting operations in these forests, we recommend a strategy which maintains a variety of approaches to management. Such a strategy is akin to ‘risk spreading’ and would involve implementing a range of tree retention strategies within cutting areas as well as the adoption of a number of strategies at the landscape level (e.g. wildlife corridors).