Previous research [K. M. Dewar, L. L. Cuddy, and D. J. K. Mewhort, “Human Learning and Memory,” J. Exp. Psych.3, 60–67 (1977)] has shown that the pitch of a single tone from a tone sequence was recognized more accurately when recognition was tested within the context of the original sequence that when it was tested in isolation. Relational factors, e.g., pitch contour and interval coding, may operate to help overcome the basic limitations on processing the pitch of isolated tones. The role of pitch versus temporal context and the effect of contour changes were examined. Musically trained and untrained listeners were presented with partial‐context and full‐context conditions (where they were given some information about the pitches surrounding the test pitch) along with a temporal‐context condition (where they were given information only about the temporal position of the test tone). Random seven‐tone sequences generated by a PDP‐11 computer were used, each tone being presented for one‐half second. For all listeners, performance was better in the partial and full‐context conditions than in the temporal‐context condition. Listeners made significantly more errors when the contour (the pattern of ups and downs) of the test sequence remained the same as the standard that when it changed, regardless of the distance between the correct and incorrect pitches. Listeners did make fewer errors the father apart the pitches. [Work supported by the Canadian National Research Council.]