Passerines maintain low levels of fat in winter even though larger fat reserves would provide better insurance against starvation. This is believed to be a result of predation risk and/or metabolic expenditure increasing with the amount of fat carried. Recent empirical studies indicate that the effect of increased mass on predation risk is small, but the effect on metabolic expenditure is large. Using dynamic modelling, I investigated how mass-dependent costs affect the pattern of fat gain. I found that increases in metabolic expenditure were sufficient to explain strong regulation of the level of fat, but that increases in direct predation risk cannot be excluded. A plausible explanation for the increase in metabolic expenditure is if the extra weight is compensated for by a parallel gain in flight muscle. Such compensation means that an increase in instantaneous predation risk can be avoided, but that costs depending on an increase in energy intake may occur instead. For example more time spent foraging increases exposure to predators, Hence, one form of predation risk (impaired ability to escape from an attacking predator) may be traded for another (more time spent foraging). This will be beneficial if this extended foraging is not risky, whereas failure to compensate for the extra mass would mean the risk of being caught by a predator increasing considerably. Besides the widely recognized trade-off between starvation and predation, this means that there may also be a trade-off between different forms of predation in wintering birds. (C) 2001 The Association for the Study of Animal Behaviour.