One hundred thirty-six rats were trained while water-deprived to lever-press for water reinforcement. Then, with the lever retracted and while food-deprived, different groups were administered either 40, 80, 160, or 320 pairings of a neutral stimulus (s 1 *) compound, light and feeder "click," with primary reinforcement (s R ). The 45mg reward was presented on either a 25, 50, or 100 per cent schedule. The design was thus a 3 X 4 factorial. Control groups which received no pairings (N = 10) or 80 non-reinforced presentations of the s" (N = 6) were included. During testing, with the lever inserted, the light and "click" compound followed each lever-pressing response. The results indicated that effectiveness of secondary reinforcement was an inverted U-shaped function of percentage of s R and an over-all increasing linear function of number of s N -s R pairings.