Between 2 and 11 months of age, the risk of sudden infant death syndrome (SIDS) declines more slowly in black infants than in infants of other races. This phenomenon might also be a feature of certain non-SIDS causes of death. Identifying these causes may through analogy provide support for the theory that SIDS is a disease of the central nervous system, an unusual consequence of respiratory infection, or a form of suffocation. We used logistic regression analysis on details of infant deaths in the United States, 1985-1991, to examine the difference between the rates of decline with increasing age in the mortality rates of black infants and infants of other races. We defined slower rate of decline in black infants as a positive difference. The magnitude and direction (positive) of the difference for deaths due to respiratory infection were similar to those for SIDS. It is unlikely that this difference in the rates of decline for respiratory infection can be explained by diagnostic cross-misclassification between respiratory infection and SIDS. SIDS appears to be a disease of the respiratory system caused by infection that affects that system's control centers.