In an earlier communication (Aboul-Dahab et al. 1958) the authors reported on the variability of response to chloromycetin of infants admitted to hospital (August 1954-1955) during a major study on gastroenteritis (Tawil et al., 1959). A relevant feature was the high incidence of relapses among the inpatients of period 3 (May 28, 1955, to August 15, 1955) of that study, particularly the chloromycetin-treated group of cases. The relapses usually coincided with the recovery, from the stools, of strains of Escherichia coli that were resistant to chloromycetin. These strains possessed the same somatic 0 antigen and Dr. F. Qrskov, who kindly checked our findings,