Daniel Kahneman and his colleagues have made an interesting discovery about people's preferences. In several experiments, subjects underwent two separate ordeals of pain, identical except that one ended with an added amount of diminishing pain. When asked to evaluate these episodes after experiencing both, subjects generally preferred the longer episode--even though it had a greater objective quantity of pain. These data raise an ethical question about whether to respect such preferences when acting on another's behalf. John Broome thinks that it is wrong to add extra pain in order to satisfy a person's preference for a better ending. His explanation for this intuition is that pain is intrinsically bad. I argue against this explanation, and raise several doubts about the moral intuition Broome endorses. In doing so, I offer alternate interpretations of Kahneman's data, and show that these each yield different values which are relevant to the ethical question .