The mechanism of spread of an epidemic of variola minor (alastrim) occurring in 1956, in Vila Guarani, a semi-rural school district of the City of São Paulo, Brazil, is examined. Cases were grouped according to time and space relationships corresponding to either a household or a school class, as once variola minor was introduced into a household or a school class spread was independent of occurrences in neighbouring social units. The constant occurrence of a single introduction in school classes and households implied that the chance for transmission of the disease outside the homes or the school was very small. The secondary attack rate was higher in households than in school classes. Twenty households with cases had more pre-school children and school children and a lower vaccination level than twenty-two neighbouring households without cases but also including one or more persons who were known to have had personal contacts with patients.
Examination of the chain of contagion clearly shows that spread in the school was directly and indirectly responsible for the majority of cases in the epidemic, for the peak of the epidemic curve and for the extent of topographical spread of the epidemic. This predominant role of the school was accentuated by incomplete preventive measures (vaccination campaign and isolation of patients) applied late in the epidemic and it is attributed to the following facts, (a) There was a single school in the district, and school children were thus concentrated, (b) The school was the only place of the district where persons living in different neighbourhoods had regular, repeated contacts, (c) The great majority of school children had had no previous variola and many had no vaccination scar, (d) The social activities of school children gave them the highest potential as transmitters of variola minor.