The direction of progress for a tree improvement program is determined by a formerly defined objective, that is the linear combination of traits to improve multiplied by their respective economic weights. These weights are derived from production functions, which simply describe the impact of trait change on some economic measure (usually profit). In this sense, they have relevance to both silviculture and breeding. Nevertheless, often objectives in forestry have either not been quantified or have been imprecisely defined, being more the expression of wishful thinking than the product of economic analysis. The CRC for Sustainable Production Forestry (CRC-SPF) has been at the forefront of the development of such objectives for vertically integrated Eucalyptus globulus and Pinus radiata industries. We review basic theory, the current status of breeding objectives development and explore several issues that hinder their usefulness. Some of the problems affect the construction of the objective function (e.g. poor knowledge of the relationship between biological traits and end-product properties, uncertainty on both cost structure and future market prices), other problems affect the application of the objective in the selection process (e.g. lack of good estimates of relationship between selection criteria and objective traits) and, finally, some problems impact on transforming gain in the objective into economic advantage for the industry (e.g. poor market signals between layers of the industry). We discuss these problems and suggest ways to overcome them and/or flag them as areas for further research. It seems to be that the major obstacle for developing formal breeding objectives is uncertainty. We postulate that breeders already have enough information to develop and use sensible breeding objectives.