A series of experiments investigated the finding that the fundamental frequency (F0) discrimination of a group of unresolved harmonics (the "target") is impaired more by a masker which starts before and ends after it than when the masker and target are gated on and off together [Carlyon, J. Acoust. Soc. Am. 99, 517-524 (1996)]. Generally, the masker was a group of unresolved harmonics with an F0 of 210 Hz, the target F0s were geometrically centered on 210 Hz, both the targets and the masker were filtered between 3900 and 5400 Hz, and the target duration was 200 ms. The additional deterioration produced by the portion of the masker occurring before the target (the "forward fringe") was greater than that produced by the portion after the target ("backward fringe"), but both had some effect. The forward and backward fringes reduced sensitivity even when the portion of the masker synchronous with the targets was absent, although the reduction was greatest when it was present. The deterioration was markedly reduced by filtering the fringes into a frequency region remote from the target and the synchronous portion of the masker, by attenuating the fringes by 10 dB, or by presenting them either contralaterally to the target or diotically. It could not be reduced by allowing the F0 of the fringes and of the synchronous portion of the masker to differ greatly from that of the target, or by adding a low-frequency portion to the fringe, thereby providing an additional cue to the time of transition between fringe and target. Explanations based on peripheral adaptation and on seemingly similar effects previously observed with brief targets [Massaro, J. Acoust. Soc. Am. 58, 1059-1065 (1975); Kelly and Watson, J. Acoust. Soc. Am. 79, 1934-1938 (1986); Divenyi and Hirsch, Percept. Psychophys. 17, 246-252 (1975)] were rejected. The data are discussed in terms of a central mechanism which includes parts of the fringes in its estimate of the pitch of the target/masker mixture.