The use of anaerobic incubation for the culture of Streptococcus pneumoniae from sputum was compared with incubation in carbon dioxide in air. A coagglutination test for pneumococcal antigen was used as an index of the number of specimens containing pneumococci. A total of 334 specimens were examined. There was evidence of pneumococcal colonisation by culture or coagglutination, or both, in 48 (14.37%), of which 41 (12.27%) yielded S pneumoniae on culture. Anaerobic incubation was better than incubation in carbon dioxide in air for the primary culture of S pneumoniae from sputum. Primary isolation of S pneumoniae was achieved in 11 of the 41 strains (26.82%) by anaerobic incubation alone, by incubation only in carbon dioxide in air in one strain (2.43%), and by both anaerobic incubation and incubation in carbon dioxide in air in 29 strains (70.73%). Anaerobic incubation gave large moist or mucoid colonies that were easy to recognise, but it suppressed the typical draughtsman colony of S pneumoniae. The factor V supplement routinely used in our medium also inhibited the formation of draughtsman colonies. It is suggested that draughtsman colonies occur because of a relative lack of the coenzyme nicotinamide adenine dinucleotide (factor V), which is required as a reducing agent in aspartate and glutamate metabolism. This nutritional deficiency may lead to bacterial cell wall defect and hence to the autolysis which gives the typical draughtsman colony.