Cell fractions have been obtained from rat livers after 6 days' starvation, and at frequent intervals during the first week of refeeding. The livers were weighed, perfused, homogenized in 0.88 M sucrose, and separated by differential centrifugation into a nuclear, two mitochondrial, a microsomal, and a final supernatant fluid fraction. The fractions were analyzed for protein-nitrogen and nucleic acids, and the results were expressed as the quantities in the average cell and the average whole liver. In agreement with the results of other workers, there was no significant alteration in the DNA or cell content of these livers, and therefore, the two methods of expressing the results are comparable. Three-hour incorporation of inorganic-phosphate-P 32 into the nucleic acids, and of glycine-2-C 14 into the nucleic acids and protein of each fraction were also measured. The biochemical findings were correlated with the microscopical appearance of the parenchymal liver cells. After 6 days of starvation, the livers lost about 50 per cent of their weight, with a corresponding reduction in the total protein and RNA contents. The losses of protein and RNA were proportional in the various cell fractions, except for a relatively smaller loss in the nuclear fraction, disproportionately large loss in the small mitochondria fraction. On refeeding, the weight of the liver increased about 50 per cent on the first day. In contrast, the early rise in the total protein and RNA was small. The earliest and most striking response occurred in the small mitochondria fraction, in which the protein and RNA content doubled in the first 24 hours. Early but less dramatic increases were also observed in the nuclear and microsomal fractions. Later, the protein and RNA of all fractions increased rapidly, approaching normal levels after about 7 days of refeeding. At different times during refeeding, in the several experiments described here, both the protein and RNA of the nuclear fraction increased abruptly to about twice the initial quantities and then decreased rapidly to the former levels. The increase in the quantity of protein in the nuclear fraction was accompanied by an increase in the specific activity and total labeling of the protein in this fraction. During refeeding, the cytoplasm as a whole accumulated protein in a straight line fashion except for a sudden increase which corresponded exactly in quantity and timing with the decrease in the protein content of the nuclear fraction mentioned above. Cytological study of the nuclei during these experiments showed that the mean nuclear volume followed the course of changes observed in the protein content of the nuclear fraction, and that at the time when the protein of the nuclear fraction was at a maximum, the nuclei were large and round, with a smooth surface, while at the time when the protein of the nuclear fraction had decreased abruptly, the majority of the nuclei were smaller, with a wrinkled surface. An appreciable number of the wrinkled nuclei were indented around large vacuoles. The total RNA content of the liver increased only during periods of active change in the protein content of the nuclear fraction. Considered together, these results suggest that in rat liver, during recovery from starvation, protein is synthesized by two processes: one is an intermittent production of protein by the nucleus, this process being accompanied by a net synthesis of RNA, and the other is a steady production of protein in the cytoplasm, independent of the fluctuations in the metabolic activity of the nucleus.