Three experiments on the recognition of short melodies investigated the influence of contour and interval information (respectively, the pattern of changes in pitch direction and the ordered sequence of pitch distances in a melody). Subjects rated pairs of melodies as "same" or "different" on a five-point scale. Six conditions were defined by two delays (short, 1 sec; and long, 30 sec) and three item types (target, related, and lure). In Target pairs, the second melody retained the contour and interval information of the first melody, being an exact transposition to another key. In Related pairs, only the contour information was retained, while in the Lure pairs neither contour nor interval information was retained. In conformity with the reports of Dowling and Bartlett (1981), the results indicated that contour information had a larger influence on recognition at short delays, whereas interval information had a relatively larger influence at long delays. The results are also consistent with an alternative interpretation stressing the importance of tonality/modality information in melody recognition at long delays.