In 90 patients referred to the pulmonary function laboratory for evaluation of hyperventilation syndrome (HVS) and in whom somatic causes of the complaints had been excluded, we investigated the degree of concordance between three widely applied diagnostic methods: 1) the standardized Nijmegen questionnaire on major daily complaints; 2) the reproduction of the same symptoms during the hyperventilation provocation test (HVPT); and 3) the responses of end-tidal CO2 fraction (FETCO2) during the HVPT. In 86% of the patients a concordance was found between Nijmegen questionnaire (i.e. score of 24/64 or more) and symptom reproduction during HVPT (i.e. recognition of at least 2 major daily complaints). Based on these combined data we made a definite diagnosis of HVS in 37 patients, of non-HVS in 40 patients, and we retained only a possible HVS in 13 patients. Each of the 16 complaints in the questionnaire contributed significantly to the distinction between HVS and non-HVS patients, and a striking similarity in rank order of daily complaints and reproduced symptoms was found. The responses in FETCO, during HVPT had little additional diagnostic value. A spontaneous fall of at least 0.25% FETCO2 during the 5 min adaptation period before the HVPT, was most reliably correlated with the aforementioned diagnostic criteria of HVS: specificity 83%, sensitivity 57%, and accuracy 70%. Neither the 3 min FETCO2 ratio nor the 5 min FETCO2 ratio during recovery after the HVPT showed a good correlation with the other diagnostic criteria. The response of FETCO2 during HVPT did especially not provide additional useful diagnostic information in the 13 patients with only possible HVS.