Selective decontamination of the digestive tract (SDD) has been evaluated as a method to prevent colonization and infection in ventilated patients in 40 trials. On the basis of an assumption that cross-infection would be reduced as a consequence of SDD and that this would distort the results of SDD studies that used concurrent controls, 14 studies used historic controls. To test this assumption, three observations from the two types of studies were compared. (i) The differences between observed and expected event rates for each study were used to perform a meta-analysis. This revealed that the summary odds ratios for bacteremia and respiratory infection were marked by significant heterogeneity (P > 0.95) and inconsistencies between those derived from studies with concurrent versus studies with historic controls. (ii) Where the data were available, the rates of acquisition of colonization in control groups were higher in studies with concurrent controls than in studies with historic controls. (iii) At least four studies with concurrent controls have shown a pattern of pathogenic isolates consistent with cross-infection between groups. These results are contrary to the initial assumption and suggest the possibility that SDD represents a major cross-infection hazard.