This study seeks to determine whether newborns are sensitive to an operant-conditioning task involving an unprepared relation between a response and a stimulus. The High-Amplitude Sucking procedure, which is based on such a relation by reinforcing nonnutritive sucking with auditory stimulation, was used. In order to verify that newborns learn the contingency between sucks and sounds in the HAS paradigm, three experiments were carried out. In Experiment 1, the effect of contingent versus noncontingent presentation of speech sounds on newborns' sucking activity was investigated during the minutes following a silent baseline. In contrast to what has been reported with 2-month-old infants in HAS, contingently stimulated newborns did not differ significantly from a nonstimulated control group. Experiment 2 showed that an increase in sucking rates could be obtained after a stimulus change, when sounds were presented contingently, but not when sounds were presented noncontingently. Experiment 3 demonstrated that newborns' sucking responses were reinforced by variation in the presented speech sounds. The results of these three experiments support the hypothesis that newborns tested under the HAS procedure are involved in an operant-learning situation. Implications for learning in neonates and possible differences with older infants are discussed.