1. In freshly isolated rat hepatocytes, the activity of the AMP-activated protein kinase is high, but decreases by 5-10-fold during incubation of the cells for 60 min. The expressed activity of acetyl-CoA carboxylase is initially very low, then rises in a reciprocal manner to the AMP-activated protein kinase activity. For both enzymes, treatment of partially purified preparations under dephosphorylating conditions abolishes the difference in activity between freshly isolated and preincubated cells. Thus, both the high activity of the AMP-activated protein kinase and the low activity of acetyl-CoA carboxylase in freshly isolated cells can be explained by phosphorylation. 2. Immediately after isolation, the hepatocytes have AMP/ATP ratios that are unphysiologically high (approximately 1:1.5). During incubation of the cells for 60 min, AMP levels fall and ATP levels rise so that the ratio becomes about 1:15, close to previous estimates of the ratio in freeze-clamped liver. The fall in AMP/ATP ratio precedes the decrease in AMP-activated protein kinase activity. 3. In cells which have been incubated for 60 min, treatment with 20 mM fructose, which causes a large but transient increase in the AMP/ATP ratio, also causes concomitant activation of the AMP-activated protein kinase and inactivation of acetyl-CoA carboxylase. 4. In all cases described above, the increases in activity of acetyl-CoA carboxylase were blocked by treatment with the cell-permeable protein phosphatase inhibitor, okadaic acid. However, the decreases in activity of the AMP-activated protein kinase were not blocked by this inhibitor. This is consistent with the finding that okadaic-acid-insensitive protein phosphatase 2C is the most effective at dephosphorylating the kinase in cell-free assays. 5. The results above suggested that AMP either promotes phosphorylation, or inhibits dephosphorylation, of the kinase. Studies in a partially purified cell-free system suggested that the former hypothesis was correct; reactivation of dephosphorylated AMP-activated protein kinase by kinase kinase was completely dependent on the presence of AMP. 6. Our results, obtained in both intact cells and a cell-free system, suggest that rises in the AMP/ATP ratio promote phosphorylation of the AMP-activated protein kinase by the kinase kinase, as well as causing direct allosteric activation. This represents a very sensitive system for switching off lipid biosynthetic pathways when ATP levels are limiting. The results with okadaic acid also suggest that protein phosphatase 2C is mainly responsible for dephosphorylation of the AMP-activated protein kinase in intact hepatocytes.