Problem-solving tasks are classified on the basis of four task characteristics (ambiguity, number of solutions, complexity, and experience), three process categories (preparation, production, and evaluation), and four general outcome categories (number of solutions, time to solution, quality, and process measures). The classification scheme is compared with recent attempts to define "ideal" characteristics of measures, and seven general criteria are proposed for evaluating problem-solving tasks. Twelve categories of tasks commonly used in research are classified and evaluated. Concept identification, switchlight problems, verbal mazes, and simulations are concluded to be tasks that most closely satisfy all evaluation criteria, although each has some problems. Implications for improvement of measures of problem-solving are indicated.