To examine the possibility of response bias, the 118 LBW subjects who produced satisfactory spirometric test results were compared with the remaining 69 LBW subjects in the cohort (excluding the 57 individuals who are known to have died). The mothers of subjects with satisfactory test results had a higher family income at birth (family income less than the minimum wage per month, 32.2% vs 15.7%, respectively, p = 0.02), higher mean weight at the beginning of pregnancy (54.4 kg vs 50.9 kg, respectively, p = 0.02), higher age (26.4 years vs 23.2 years, respectively, p = 0.001), and lower prevalence of smoking during pregnancy (49.2% vs 65.2%, respectively, p = 0.03). The adolescents had higher mean weight (63.6 kg vs 55.9 kg, respectively, p = 0.03) and height (171.4 cm vs 166.6 cm, respectively, p = 0.02) in 2000. There were no significant differences in terms of maternal education, height of mothers, gestational age, birth weight, schooling, and height and smoking of adolescents. Some degree of response bias is unavoidable in long-term cohort studies, and this has to be taken into account when interpreting the present results. To produce bias in the results, a factor must be associated not only with the rate of follow-up but also with the outcome under study. Of the above-listed variables associated with birth weight, family income, weight at the beginning of pregnancy, maternal age, smoking, and height of 18-year-old subjects were also associated with lung function. However, all of these variables were included as confounders in the multivariable analysis and therefore were taken into account when comparing LBW with other subjects.