More than 500, or 97% of all 3-year-old children in a Danish municipality, were seen in January 1976 for middle ear examination, including tympanometry. Persistently abnormal cases were checked at the same intervals during the subsequent 6 months, without treatment. Thus, the type and duration of impaired middle ear function in otherwise unselected 3-year olds were mapped. In order to investigate the long-term significance of middle ear pathology of various description in early childhood, the children were retested exactly 3 years later. The study revealed that even a most extensive programme of serial tympanometry at the age of 3 years could not distinguish all children who would have poor middle ear function and hearing loss at 6 years of age from the total group of normals. Therefore, a programme of screening 3-year olds by impedance tests cannot replace existing school screenings. Cases showing flat curves with gradients ≤ 0.1 at each retest during 3 months made up consistent middle ear pathology over a long period of time, giving the poorest prognosis: about 55% went on having a flat curve, and about 40% did not pass a puretone sweep test at the age of 6 years. 3-year olds with flat curves and a favorable long-term prognosis revealed themselves directly by showing a great test-retest instability of the tympanometric pattern. Even prolonged shifting between flat curves and a pronounced negative middle ear pressure during preschool age did not entail poor long-term prospect. The influence of these findings upon rational tightrope walking between expectation and therapeutic intervention in clinical pediatric otology and upon preschool impedance screening is outlined.