The paper records the results of a retrospective study of laboratory findings on samples sent for diagnosis of possible tuberculosis among patients attending or admitted to hospital during the 8 years, 1968-1975. Out of a total of 23 419 such samples, 2479 were positive for tubercle bacilli: both direct microscopy and cultural examinations were carried out, except in the case of many sputum samples which were strongly positive on direct film examination. It was found that the simple examination of stained smears would detect 77% of positive sputa, but was unsatisfactory for other specimens, where cultural examination was essential. Among the positive samples, only sixty-nine came from children under 15 years old: this low figure probably represents selection in hospital attendance or submission of samples, rather than a true comparison with the incidence of clinical tuberculosis in adults. The type of infections evident in children differed from that apparent for all age groups combined, in the high proportion of cases of tuberculous meningitis among the younger age group. The yearly figures give little indication of a fall in the incidence of clinically significant tuberculosis over the past 8 years. A vigorous effort is required for case finding and treatment if there is to be substantial improvement over the next 8 years.