AIMS--To use the polymerase chain reaction (PCR) to detect Chlamydia pneumoniae and Chlamydia psittaci in sputum samples. METHODS--A nested PCR was developed, the first stage of which amplified DNA from both C pneumoniae and C psittaci while the second stage targeted specifically at C pneumoniae, allowing the two species to be differentiated. The primers were designed not to amplify sequences from C trachomatis. A panel of 26 sputum samples from patients with community acquired pneumonia evaluated previously by enzyme linked immunosorbent assay (ELISA), direct immunofluorescence (DIF), and culture was tested blind by PCR. Most of these specimens also had accompanying serial serum samples which were tested for species specific antibodies using microimmunofluorescence (micro-IF). RESULTS--PCR detected C pneumoniae DNA in 10 of the 26 samples and C psittaci DNA in four. There was good concordance between ELISA, DIF, micro-IF and PCR in the C pneumoniae group. Two of the C psittaci identified by PCR were labelled C pneumoniae by DIF but the PCR results were supported by serology or a history of bird contact. Of the PCR negative group: six were true negative results; two contained C trachomatis. There were four discrepant results. CONCLUSIONS--The data suggest that PCR is effective in the detection of C pneumoniae. The sensitivity for C psittaci is inevitably lower due to the strategy taken but specificity seemed to be good.