Blood culture isolates of Candida albicans were collected from 102 patients in Seattle, Wash., hospitals (n = 77) and Hong Kong (n = 25). The patients were classified by immune status into two groups. Group I patients were severely immunosppression, and group II patients had underlying risk factors for candidemia but no underlying immunosuppression. Serotyping by Hasenclever tube agglutination was done. In the Seattle area, the odds of fungemia with type B C. albicans were 3.62 times greater than the odds of type B fungemia in group II patients. Although the odds ratio could not be computed for Hong Kong patients, the direction of the relationship in this population was consistent with the data on Seattle patients. Despite the magnitude of the odds ratios, the relative prevalence of type B over type A in group I compared with group II was not significant when analyzed separately by region, probably because of relatively low numbers of isolates in group II. Accepting that the effect of immune status on serotype is equivalent across regions but presupposing that a regional effect on type B prevalence exists, the pooled odds for fungemia with serotype B in group I patients are increased 5.4-fold over those of group II patients. Logistic regression analysis controlling for region gave similar results.