Normative cut-off values for the bronchodilator reversibility test are published neglecting factors that may influence the test result other than disease. The objective of this cross-sectional study of a general population was to examine how a salbutamol reversibility test response is depending on anthropometrical variables and smoking. An age and gender stratified random sample of all adults aged 47-48 and 71-73 years living in Bergen, Norway, was invited. The 3506 attendants (69%) filled in a questionnaire and performed spirometry before and after inhalation of 400 microg salbutamol. The mean (SD) absolute FEV(1) bronchodilator response, the change in % predicted, and the change in % initial among middle-aged were 71 (122) ml, 2.0 (3.3)%, and 2.4 (4.1)%, and in the elderly 64 (113) ml, 2.4 (4.3)% and 3.3 (5.9)%, respectively. In a multiple linear regression analysis including adjustment for the initial FEV1 in % predicted, smoking and pack-years were negatively correlated to all indices. Current smoking was considerably more important than past smoking. Female gender, old age, and BMI were positively correlated with the percentage change indices, but the correlation with BMI decreased with increasing heights. These trends were unchanged after excluding subjects with obstructive lung disease or coronary heart disease. This study demonstrates that smoking habits predict all indices expressing the salbutamol bronchodilator response among middle-aged and elderly from a general population. Also, the change in % predicted and the change in % initial indices are dependent on anthropometrical variables. However, although smoking and anthropometrical variables, as well as level of lung function predict the response to inhaled beta2-agonists, these factors explain only 7-16% of the total variation of the measurement indices, and seem therefore of minor importance to the interpretation of the test result.