Chronic obstructive pulmonary disease (COPD) is a progressive disease with alveolar destruction (emphysema) and bronchiolar fibrosis (obstructive bronchitis) in variable proportions. Reducing disease progression, as assessed by forced expiratory volume in I second (FEV1) decline, health-related quality of life, exacerbation rate and mortality, is a more realistic outcome than physiological improvement. This paper reviews all the published studies of at least 100 patients followed for at least 2 years. Studies have included patients with mild COPD (Copenhagen City Lung Study) to advanced symptomatic disease [Inhaled Steroids in Obstructive Lung Disease (ISOLDE)], with 2 studies of those with relatively early symptoms [European Respiratory Society Study on Chronic Obstructive Pulmonary Disease (EUROSCOP) and Lung Health-21. Exacerbation frequency, and probably severity, are reduced by high dose inhaled corticosteroids. Exacerbations are only frequent in more advanced disease, limiting the use of this outcome in EUROSCOP and Lung Health-2. Exacerbations are associated with reduced health-related quality of life. ISOLDE clearly showed a reduced rate in decline of the disease-specific St George's Respiratory Questionnaire with fluticasone propionate, partly related to the reduced exacerbations. The symptom component of the score showed the greatest difference between placebo and fluticasone propionate. None of the larger studies were able to reproduce the statistically significant reduction in the rate of decline in FEV1 suggested by the smaller, earlier studies. This might at least in part be as a result of the statistical modelling used which cannot adequately compensate for those with more rapidly progressive disease dropping out earlier. The equivalent doses of inhaled corticosteroids differed approximately fivefold between the major studies. The more positive results were obtained with higher doses. Oropharyngeal adverse effects were similar to those seen in patients with asthma; bruising was increased in one study with budesonide, otherwise adverse effects were similar to placebo. Bone loss was specifically studied in subgroups of patients in EUROSCOP and Lung Health-2. Budesonide 800 microg/day was associated with less bone loss than placebo, whereas triamcinolone 1200 microg/day was associated with excess bone loss. High dose inhaled corticosteroids have a favourable risk/benefit ratio in patients with advanced disease, particularly those with frequent exacerbations, and no benefit for those with very mild disease. It is not possible from the data to make firm recommendations for the important intermediate group where delaying progression is likely to lead to greatest benefit. I believe high dose inhaled steroids are warranted for those with intermediate severity COPD, who have frequent exacerbations or significant COPD-related symptoms.