The objective of this study was to compare the two most frequently used asthma-specific quality of life (QOL) questionnaires and two generic QOL questionnaries with clinical data in order to establish their ability to discriminate in asthma severity, also referred to as cross-sectional construct validity. We conducted a multicentre, randomized, placebo-controlled study investigating the long-term effects of the long-acting beta 2-agonist formoterol in asthma patients using inhaled corticosteroids and short-acting beta 2-agonists. Before randomization, the peak expiratory flow (PEF) and use of beta 2-agonists for relief of symptoms were recorded twice daily for 1 month. At the end of the run-in period, the forced expiratory volume in 1 s (FEV1), airway hyper-responsiveness (PC20 methacholine) and QOL were measured, using two asthma-specific questionnaires, the Asthma Quality of Life Questionnaire (AQLQ) and the Living with Asthma Quality of Life Questionnaire (LWAQ), as well as two generic questionnaires, the Short Form 36 (SF-36) and the Psychological General Well Being (PGWB) index. The quality of life of the patients (n = 110) was only mildly impaired. The mean symptom score was 3.6 on a scale of 0-21 and the mean FEV1 was 65% of that predicted. The correlations between the QOL parameters and objective measures of asthma severity, such as the FEV1, PEF and PC20, were low (0.01-0.37). The correlations between the QOL parameters and subjective measures of asthma severity, such as symptom scores (range 0.26-0.65) and beta 2-agonist use for the relief of symptoms (0.09-0.39), were higher. The correlations of the activity and symptoms domains of the AQLQ with diary-obtained symptom scores were r = 0.50 (p < 0.0001) and r = 0.65 (p < 0.0001), respectively. The correlation between the physical construct of the LWAQ and symptom scores was r = 0.42 (p < 0.001) and that between the physical sumscore of the SF-36 and symptom scores was r = 0.50 (p < 0.001). The correlation between the PGWB and symptom scores was r = 0.41 (p < 0.001). The relation between the FEV1, PEF and PC20 and QOL was very low. The QOL measurements related well to the symptom scores. The AQLQ had a better cross-sectional construct validity than the LWAQ. The cross-sectional construct validity of the SF-36 was surprisingly high for this group of patients and the SF-36 performed better than the PGWB and LWAQ.