OBJECTIVES: To review the clinical effectiveness and cost-effectiveness of a new technology, the inhaled insulin, Exubera (Pfizer and Sanofi-Aventis, in collaboration with Nektar Therapeutics), a short-acting insulin. DATA SOURCES: Electronic databases were searched up to November 2005. REVIEW METHODS: A systematic literature review was conducted and economic modelling carried out. An industry model was used for modelling. RESULTS: Nine trials of inhaled insulins were found, but only seven used the Exubera form of inhaled insulin. The other two used inhaled insulins that have not yet been licensed. There were five trials in type 1 and two in type 2 diabetes. Inhaled insulin is clinically effective, and is as good as short-acting soluble insulin in controlling blood glucose, plus it works slightly more quickly. None of the published trials compared it with short-acting analogues. Most patients in the trials were on combinations of short-acting, and either long- or intermediate-acting insulin, and both were changed, making it more difficult to assess the effects of only the change from soluble to inhaled insulin. Patient preference was the only significant difference between inhaled and soluble insulin in the trials. Most patients preferred inhaled to injected short-acting insulin, and this has some effect on quality of life measures. However, the control groups mostly used syringes and needles, rather than pens. As pens are more convenient, their use might have narrowed the patient satisfaction difference. There were no trials of inhaled insulin against continuous subcutaneous insulin infusion (CSII). No serious adverse experiences of inhaled insulin in the lung have been seen to date, but it is too soon yet to judge long-term effects. The manufacturer's model appears to be a high-quality one, although the results depend more on the assumptions fed into the model than on the model itself. The key assumptions are the size of the gain in quality of life utility from inhaling rather than injecting insulin, the effect of having an inhaled option on the willingness to start insulin among people with poor diabetic control on oral drugs, and the effect on glycaemic control. We consider that these assumptions make the cost-effectiveness appear better than it really would be. The manufacturer's submission assumed utility gains of 0.036-0.075 in patients with type 1 diabetes, and 0.027-0.067 in those with type 2, based on an unpublished utility elicitation study sponsored by the manufacturer. We thought that these gains were optimistic and that gains of 0.02 or less were more likely, on average. However, patients with particular problems with injection sites might have more to gain, although they might also be a group with much to gain from CSII. A key factor is the cost of inhaled insulin. Much more insulin has to be given by inhaler than by injection, and so the cost of inhaled insulin is much higher than injected. The extra cost depends on dosage but ranges from around 600 pounds to over 1000 pounds per patient per year. CONCLUSIONS: The inhaled insulin, Exubera, appears to be as effective, but no better than injected short-acting insulin. The additional cost is so much more that it is unlikely to be cost-effective. The long-term safety is uncertain. Additional research is recommended into the safety, efficacy and cost-effectiveness of inhaled insulin.