A prospective study was initiated before the expected rubella epidemics in 1964 and 1965 in Los Angeles. Seventy-six families were evaluated by means of rubella complement fixing (cf) antibodies. The cf test, which has notable limitations, was chosen as a serologic test because it was possible to secure repeated samples of sera from all members of the families if venipuncture could be avoided. Definite evidence of clinical or serological rubella occurred in 13 of 399 persons enrolled in 1964, an attack rate of 3.3 percent. Four persons had clinical rubella only, five had clinical disease with seroconversion and four had seroconversion only. The ratio of apparent to unapparent disease was nine to four. There were four key families, each of which had more than one individual with definite clinical or serological evidence of rubella, suggesting that clustering of rubella cases does occur in families having an index case. In these families three types of intra-family spread were demonstrated: (1) all affected members had clinical disease, (2) all those affected had only inapparent disease, and (3) both apparent and inapparent disease in the same family.