The Registrar General's decennial supplements on occupational mortality provide only limited information on mortality in the armed forces in the United Kingdom. Mortality has therefore been studied among a group of 30,619 United Kingdom servicemen who served abroad in tropical or desert areas in the 1950s and 1960s, and who remained in the services for a total of at least five years. Mortality from all causes of death, all neoplasms, and all other known non-violent causes was lower than that expected from rates for all men in England and Wales, whereas mortality from accidents and violence was raised. These differences remained after adjustment for social class, affected both officers and other ranks, and had not disappeared even after the men had been followed up for at least 20 years. When mortality from 20 specific cancers and 10 other disease groups was examined there were significant excesses for cancers of the oesophagus (standardised mortality ratio (SMR) = 146; p = 0.03) and prostate (SMR = 156; p = 0.03), and significant deficits for cancers of the lung (SMR = 73; p less than 0.001), stomach (SMR = 66; p = 0.002), bladder (SMR = 53; p = 0.02), other specified neoplasms (SMR = 48; p = 0.001), coronary heart disease (SMR = 76; p less than 0.001), bronchitis, emphysema, and chronic obstructive lung disease (SMR = 42; p less than 0.001), and for five further groups of diseases unrelated to smoking or alcohol. Examination of mortality in each of the three services separately identified two specific hazards in the Royal Navy; seven deaths from mesothelioma occurred compared with less than 2.06 expected (p less than 0.005), and there was also an excess of neoplasms and of other diseases associated with alcohol (SMRs of 181 and 229; p = 0.002 and less than 0.001). Mortality from smoking related diseases other than those associated with alcohol was low in all three services, particularly among officers.