Many consider that the compliance of the middle ear as measured from the tympanogram can be helpful in diagnosing otosclerosis. To test this assertion, the compliance in 34 individuals with surgically proven otosclerosis was compared with the compliance in 34 age and sex matched, normal controls, randomly selected from the population. Though the mean compliance was different in the two groups, there was considerable overlap in the range of values which severely limits the practical usefulness of tympanometry. If the level of compliance is taken at which a false negative diagnosis would be made in 10 per cent of otosclerotic ears, a false positive diagnosis of otosclerosis would be made in 88 per cent of normal ears. If the level of compliance is taken at which a false positive diagnosis of otosclerosis would be made in 10 per cent of normal ears, 72 per cent of ears with otosclerosis would be considered normal. It is concluded that tympanometry will not help to arrive at a diagnosis of otosclerosis.