Aims/hypothesis. Insulin control of glucose output is a major mechanism by which appropriate amounts of glucose are produced to supply energy to the central nervous system, without causing long-term increases of the plasma glucose concentration. It is hypothesised that the primary route by which insulin maintains control over glucose production is indirect and is mediated by regulation of non-esterified fatty acid release from the adipocyte. The question arises as to why evolution has chosen insulin to be secreted into the portal vein, if control of the liver is partially or primarily indirect. It is suggested that alterations in hepatic insulin clearance which attend increases in central adiposity are an important part of the compensation for insulin resistance and limit the necessity for up-regulation of insulin secretion in insulin resistance secondary to central adiposity.¶Methods. Review of research from author's group and other laboratories.¶Results. Data over the previous decade indicate that suppression of glucose output by increased insulin is a relatively slow process, much slower than the rate of binding of insulin to hepatocytes. One explanation is that insulin acts on an extrahepatic tissue, which in turn alters a signal to the liver, reducing glucose output. Additional evidence for an extrahepatic primary effect of insulin emerges from experiments in which insulin was given portally or peripherally at half the portal dose. Endogenous glucose production was related to systemic, not portal insulin, supporting the concept that the primary step in insulin's action on liver is on some other tissue, altering signalling to the liver itself. Strong correlation between plasma non-esterified fatty acids (NEFA) and liver glucose output suggests that the primary effect is on the adipocyte. The primacy of the adipocyte locus for the insulin effect included data that insulin's action on liver is prevented when plasma NEFA are maintained, as well as data showing proportional decline in glucose production and fatty acids when antilipolysis is induced by an adenosine agonist.¶Why then, from an evolutionary point of view is insulin secreted into the portal vein? Institution of central adiposity in dogs with fat feeding causes hepatic insulin resistance, at least partially due to the provision of NEFA in portal blood. The initial response to resistance is enhanced beta-cell sensitivity to glucose; a secondary compensation is, however, a substantial reduction in liver clearance, allowing for a greater proportion of secreted insulin to reach muscle, where it can more efficiently stimulate glucose utilisation.¶Conclusion/interpretation. Non-esterified fatty acids act as a signal as well as a metabolic substrate. They can regulate glucose utilisation in muscle and apparently are important signals to the liver and the beta cells as well. The importance of portal vein NEFA concentrations to the function of the liver could explain insulin resistance of the liver with central pattern obesity. [Diabetologia (2000) 43: 946–952]