An important weakness of economic models in the field of osteoporosis has been the dependence on assumptions or expert judgements rather than empirical estimates for the utility values of key health events associated with osteoporosis such as hip, vertebral, wrist fracture and established osteoporosis. This paper seeks to identify the best available utility estimates for health states associated with osteoporosis and make recommendations about their use. It is based on a systematic search of the main literature databases. Studies meeting inclusion criteria have been reviewed in terms of the appropriateness of the valuation technique, the validity of the descriptive system (if one was used), the number and type of respondents, and overall quality of the study. Twenty three estimates of health state values (HSVs) were found across the four conditions from five studies. These empirical estimates were found to differ significantly from the commonly used assumptions in economic evaluation, but with a wide variation between estimates for the same state (0.32 to 0.80 for vertebral fracture states). This variation can be partly explained by the valuation technique, health state description and the background and perspective of respondent, and leaves scope for considerable discretion that could be abused. There are also problems in using values obtained from the study populations to those in economic models and the difficulty of predicting health state values in those who avoid a fracture. The review recommends a set of health state values as part of a "reference case" for use in economic models. Due to the paucity of good quality of estimates in this area, further recommendations are made regarding the design of future studies to collect HSVs relevant to economic models.