A series of experiments on operant variability by Neuringer and colleagues (e.g., Neuringer, 1986, 2002; Page & Neuringer, 1985) have been repeatedly cited as showing that behavioral variability can be reinforced by making reinforcement contingent on it. They showed that the degree of variability in pigeons' eight-peck sequences, as measured by U values, increased with an increasing lag requirement. As a result, Page and Neuringer concluded that variability is an operant. The argument that variability should be considered to be an operant dimension of behavior, just like other reinforceable properties, such as topography, frequency, duration, force, and location is based on two premises. First, there has to be a correspondence between reinforcement contingencies that require behavioral variability and the actual variability of behavior. Second, alternative explanations of that variability need to be ruled out. There is abundant empirical evidence that shows that behavioral variability increases when reinforcement is contingent on it. Identifying the sources of behavioral variability and novelty is important in behavior analysis, and Neuringer and colleagues are to be commended for their thorough investigation of contingencies that produce predictable levels of behavioral variability. (Contains 1 figure.)