Air-borne bacteria collected from occupied schoolrooms and cultivated aerobically on serum agar yielded a mean count of' general bacterial flora' (principally various micrococci) of about 70 colonies/cu.ft. with a 95% range of 19–251 colonies/cu.ft. Of a number of other occupied places examined, none except busy public offices yielded as high an average count.When the air samples were cultivated on a medium designed to suppress micrococci, a count of 1.0–1.5 colonies of streptococci/cu.ft. was observed. About 25 % of the streptococci were levan-producing Streptococcus salivarius.An analysis was made of factors that might affect the number of bacteria found in the classroom air. The principal findings were that Str. salivarius was correlated with the amount of talking among the room occupants and that the general count was correlated with the ventilation rate and to a less extent with talking. Both counts were correlated less strongly with the general level of activity among the occupants.In classrooms with ultra-violet irradiation, the numbers of Str. salivarius were reduced and the correlation with talking was less than in the unirradiated rooms, being little greater than the small correlation with activity.It was concluded that most of the Str. salivarius in the air of the schoolrooms were derived directly from the upper respiratory tract of the occupants, and these seemed to be sensitive to ultra-violet irradiation. A smaller proportion were probably dispersed from secondary reservoirs and were less sensitive to the irradiation.Most of the organisms comprised in the general bacterial flora were probably derived from such secondary reservoirs as clothing, but some, including the enterococci, were almost certainly derived from floor dust. We are greatly indebted to Mrs Eileen Woods for her help in the computations on which this paper is based.