If you are picking berries in a field of berry patches, when is it time to move to the next patch? One obvious thought is that you should pick every ripe berry, but collecting the last berries might be more trouble that it is worth. Instead, Charnov’s (1976) “Marginal Value Theorem” (MVT) argues that the optimal time to leave comes when the current rate of return drops below the overall average rate of return. MVT successfully describes human behavior in basic human foraging tasks in the lab. For instance, if the “travel time” between berry patches is made longer, foragers will tend to stay longer in the current patch because the unprofitable travel time drives down the overall rate of return. However, there are limits to MVT. Some are obvious. If there is pressure to collect every target (e.g. in some medical settings), MVT would be the wrong definition of ‘optimal’. Others are less obvious. How should we understand foraging for rare targets where the current rate of return may drop to zero for extend periods but where we would not want the forager to give up on the patch? We will present examples of the power and limits of the MVT approach to human foraging.