A survey in Greater Boston identified 1577 new cases of histologically-confirmed Hodgkin's disease (HD) diagnosed from 1959 to 1973. The spatial position of each case was taken as the place of residence at diagnosis, and the temporal position as the month and year of diagnosis. Sufficient data for analysis were available for 1398 cases (89%). Space-time clustering was evaluated by three different methods: (a) Knox, (b) David and Barton and (c) Ederer, Myers and Mantel. When all HD cases were considered, none of these methods revealed statistically significant clustering. When stratification by risk factors was done, each of the techniques found significant clustering for various subgroups. Young adult cases (age 16-45 yr) tended to cluster more than older cases (age 46-70) and Catholics more than Jewish or Protestant cases. Rather than suggesting clustering at the time of HD diagnosis, the sporadic and inconsistently positive findings of this study may reflect case aggregation at the time of shared etiologic exposures, particularly if HD has a long and/or variable latent period.