A questionnaire concerning problems inherent to ronchopathy was evaluated in order to assess its test-retest reliability and the interobserver variability of the items. The results indicate the existence of three orders of variables. The first (class A) was characterized by good intra- and inter- observer reliability. It included all interval variables (i.e. weight, height, arterial pressure) and most ordinal variables (i.e. grading of snoring, excessive daytime somnolence, morning headache, smoking, etc.). The second (class B) was characterized by good intra-observer and poor inter-observer reliability. It included snoring onset time and morning somnolence. The third class (class C) was characterized by both poor intra- and inter-observer reliability. It included sleep apnea. For large epidemiological survey purposes the authors suggest that only class A variables be used.