C-reactive protein (CRP) has been shown to be a useful and sensitive indicator of pyogenic infections in many clinical situations, including acute pneumonia and infective pulmonary exacerbations in cystic fibrosis patients. Exacerbations of COPD are often, but not always, associated with demonstrable infection. The value of CRP measurement in this situation has not been assessed. We have evaluated CRP measurement in 50 patients [age 71 +/- 8 (SD) years] who were admitted to hospital with clinical evidence of exacerbation [PaO2 = 7.3 +/- 1.3 (SD) kPa, baseline FEV1 = 0.8 +/- 0.4 (SD) l]. These patients all had serial measurement of CRP [polarizing immunofluorescence (Abbot, TDx)], peripheral white cell count (WCC), body temperature, peak expiratory flow rate, Karnofsky performance status and chest X-ray, in addition to serial sputum bacteriological analysis carried out in a specialized laboratory. CRP was elevated (> 10 mg l-1) in all patients (n = 29) with proven infection [103 +/- 98 (SD) mg l-1]. Levels were markedly elevated in patients infected with Streptococcus pneumoniae (mean 156 mg l-1); there was also a rapid fall in the CRP with therapy. WCC fell with therapy, giving a correlation with CRP level (r = 0.44, P < 0.01). Since CRP elevation was observed in patients having exacerbation with proven infections and also in those where infection was not proven, it is possible that, while it is a marker for COPD exacerbation, it is not necessarily a marker of bacterial infection per se. However, it is evident from our study that it is of value in the assessment of exacerbations of COPD, where routine bacterial culture of sputum is often unreliable, and thus the measurement of serum CRP may provide an additional objective indicator of infection.