Pigeons' pecks on one key produced a stimulus correlated with a mixed-ratio schedule of food reinforcement. Pecks on a second key produced a stimulus correlated with a fixed-ratio schedule. When the arithmetic mean of the mixed ratios equaled the fixed ratio, the former stimulus maintained a higher rate of pecking. When the fixed ratio was sufficiently smaller, preference shifted to it. The pigeons' relative preference for the schedules could be described by comparing the geometric mean of the reinforcement rates in the several mixedratio components with the reinforcement rates in the fixed-ratio components.Chained schedules of reinforcement are a useful tool in studying secondary (or conditioned) reinforcement. In a two-link chained schedule, for example, responding in the presence of a first stimulus is reinforced by a second stimulus; responding in the presence of the second stimulus is reinforced with food (Ferster and Skinner, 1957). Kelleher and Gollub (1962) have presented evidence that responding in the early links of a chain is maintained primarily by conditioned reinforcement. If two chained schedules are concurrently available, an organism's preference for one secondary reinforcer (Sr) relative to the other Sr may be measured by the relative amount of responding (responses on one key/total responses) maintained by that Sr. The organism may be required to choose between two SrS when each is correlated with a different schedule of primary reinforcement (Autor, 1960;Reynolds, 1963). Using this technique, Herrnstein (1964a) showed that the relative rate at which pigeons' pecks were maintained by a secondary reinforcer matched the relative rate of primary reinforcement in its presence when primary reinforcement was programmed on variable-interval (VI) or variable-ratio (VR) (Herrnstein, 1964b). Instead, the pigeons responded at a higher rate when the Sr was correlated with the VI schedule of primary reinforcement. Similarly, rats were found to prefer the side of a Y-maze in which reinforcement was delayed for either 0 or 30 sec over the side in which reinforcement was always delayed by 15 sec (Pubols, 1962