A combination of "rational" and "irrational" strategies for the creation of enzymes with novel properties is proving to be a powerful concept in the field of enzyme engineering. Guided by principles of physical organic chemistry, rational design strategies are used to identify suitable target enzymes and to choose appropriate molecular biological methods for engineering purposes. In contrast, irrational (or random) strategies are centered around the biological paradigm of stochastic molecular evolution. As illustrated in this review, such a hybrid approach is particularly useful for the design of new modular enzymes.