Two experiments, one using 0 + the other 1 + rainbow trout, were conducted to investigate the effect of prolonged starvation on plasma growth hormone levels. The results from both experiments were essentially the same. As expected, starvation resulted in cessation of growth and in a lower coefficient of condition, whereas fed fish continued to grow and remained in good condition. Starvation had relatively little effect on the plasma cortisol level; in one experiment levels were elevated temporarily in starved fish, although by the end of the experiment there was no longer any difference between starved and fed fish, and in the other experiment plasma cortisol levels remained very low throughout the course of the experiment in both starved and fed fish. In contrast, in both experiments starvation had a pronounced effect on the plasma growth hormone level, which rose steadily during both experiments, such that it was six times higher after 1 month of starvation in 0 + fish, and five times higher after 6 weeks of starvation in 1 + fish. Thus, paradoxically, fed fish had very low plasma growth hormone levels and grew rapidly, whereas starved fish had elevated plasma growth hormone levels but did not grow. In both experiments a strong negative correlation was observed between the plasma growth hormone level and the coefficient of condition of the fish. The results are discussed with regard to the well-established metabolic changes that occur during starvation, and it is suggested that a major role of growth hormone during starvation is to aid in the mobilisation of fatty acids and glycerol from adipose stores.