1. Two hypotheses to account for general catabolite repression of the lactose enzymes in Escherichia coli were tested: the dilution model of Palmer & Moses (1967), and the specific catabolite repressor model of Loomis & Magasanik (1965, 1967). 2. The dilution model predicts that in mutants lacking the i-o regulation system the differential rate of beta-galactosidase synthesis should increase when amino acid-synthesizing enzymes are repressed by the presence of amino acids in the medium. It also predicts that with such mutants the total absence of P(i) from the medium should not result in the complete cessation of beta-galactosidase synthesis that is observed with wild-type cells. 3. Neither prediction was confirmed experimentally, and it is concluded that this model cannot explain catabolite repression. 4. The specific repressor hypothesis depends on the properties of a strain of E. coli carrying the CR(-) mutation. It requires both that cells of this genotype should be totally resistant to general catabolite repression and that this resistance should be specific for the lactose enzymes. 5. In fact the synthesis of beta-galactosidase by CR(-) cells, though showing resistance to catabolite repression by growth on glucose, was found to be repressed in several other circumstances. 6. Two other inducible enzymes, l-tryptophanase and d-serine deaminase, also showed resistance to repression by glucose in CR(-) cells. 7. It is concluded that this model, too, does not account for general catabolite repression. 8. Strains carrying deletions at either end of the lactose operon that extend into the structural genes of the operon continue to exhibit catabolite repression. 9. These experiments appear to eliminate the possibility that catabolite repression operates at the level of DNA transcription, and suggest that repression affects instead the translation of messenger RNA into protein.