In this paper I consider a complex decision problem where subjects have to cope with a time horizon of uncertain duration and must update their termination probabilities which depend on stochastic events during life. First I describe how economic theory suggests to solve the decision problem. But since real decision makers can hardly be expected to behave according to the theoretical solution in the problem at hand, I describe several heuristics or rules of thumb and investigate their theoretical performance. Then observed behavior and the way how people tackled the problem is described. In the second part of the paper I discuss how much of the data can be explained by assuming that experimental subjects are risk averse.