In this randomised double-blind study, patients >or=40 years old with COPD, a smoking history of >or=10 pack-years, a pre-bronchodilator FEV(1) of <or=60% predicted and an FEV(1)/FVC of <or=70% received tiotropium 5 microg or placebo via Respimat inhaler once daily for 48 weeks. Other medications were permitted except inhaled anticholinergics. Co-primary endpoints were trough FEV(1) and the time to first exacerbation. Adverse events were followed and vital status regularly assessed. In all, 3991 patients (mean age, 65 years [SD, 9 years]) were evaluable. Mean baseline FEV(1) was 1.11 L (0.40 L) or 40% (12%) of predicted normal. Adjusted mean differences in trough FEV(1) and trough FVC at Week 48 (tiotropium minus placebo) were 102 and 168 ml respectively (p < 0.0001, both). Tiotropium delayed time to first exacerbation relative to placebo (hazard ratio [HR], 0.69 [95% CI, 0.63-0.77]) and time to first hospital-treated exacerbation (HR, 0.73 [0.59-0.90]). SGRQ score at Week 48 was 2.9 units lower with tiotropium (p < 0.0001). Adverse and serious adverse events were balanced across treatment groups and similar in profile to previous tiotropium trials. The rate ratio for a major adverse cardiovascular event during the treatment period + 30 days was 1.12 (0.67-1.86). By the end of planned treatment (Day 337) 52 patients on tiotropium (incidence rate per 100 years, 2.94) and 38 on placebo (2.13) had died (HR = 1.38 [0.91-2.10]; p = 0.13). Lung function, exacerbations and quality of life were improved by tiotropium 5 microg Respimat but a numerical imbalance was seen in all-cause mortality. The protocol is registered on the European Clinical Trials Database as trial number 2006-001009-27 and in the ClinicalTrials.gov database as NCT00387088.