Rubella infections, notified by general practitioners on the basis of a clinical diagnosis, were investigated by testing blood and saliva samples for specific IgM. Overall 52 (29%) of 178 cases with appropriately timed blood specimens were confirmed as recent rubella by IgM serology. Only 2 (3%) of 74 cases in children under 5 years were confirmed compared to 50 (48%) of 104 cases in older children and adults. The confirmation rate was lower (6%) in those with documented vaccination history than in those without (42%). The specificity of saliva rubella IgM testing compared to testing corresponding blood samples was 99%. The overall sensitivity of saliva rubella IgM testing was 81%. This rose to 90% if results from inappropriately timed specimens and specimens taking more than 1 week to reach the laboratory were excluded. A corresponding saliva rubella IgG test was 98% sensitive and 100% specific. Of 126 rubella IgM negative cases, 25 (20%) were positive for parvovirus B19 IgM. This study confirmed that rubella surveillance based on clinical reports is not specific. It also demonstrated that saliva samples, if taken 7-42 days after onset of illness and transported rapidly to the laboratory, are a feasible alternative to blood samples for rubella surveillance.