Four-hundred fifty-nine sera from 342 persons were examined at Ruchill Hospital, Glasgow, using an agargrown heat-killed, ether-treated antigen derived from the Philadelphia 1 strain of the Legionnaires' disease bacterium. They were examined, in parallel, at the Center for Disease Control (CDC), Atlanta. Overall agreement (results agreeing ± one twofold dilution) was reached with 90% of sera. Important differences were obtained with 10 (2.2%) sera. With three of these 10 sera the CDC alone obtained a significant titre (256) whereas with the other seven the Glasgow laboratory alone obtained a titre of 256. Altogether 28 persons (8.2%) with an agreed titre of 256 or higher were identified. However, some of these people had been ill 4 years before serum was obtained for examination. These persistent high titres make the serologic diagnosis of Legionnaires' disease difficult.