Three experiments investigated the finding that F0 discrimination for a harmonic complex can be impaired by temporal fringes [R. P. Carlyon, J. Acoust. Soc. Am. 99, 525--533 (1996)]. The 100-ms target and the 200-ms forward and backward fringes were filtered (1375--1875 Hz) and presented in a pink noise background. Two F0's were used: 88 Hz, where the harmonics were unresolved by the peripheral auditory system, and 250 Hz, where the harmonics were unresolved. The fringe produced a larger deterioration when it had a similar F0 to the target than when the F0's differed. Experiment 2 showed that adding off-frequency components to the fringe did not affect this pattern of results, thereby ruling out interpretations based on spectral confusion between the target and the fringe. Experiment 3 further examined the dependence of the deterioration on the similarity between target and fringe F0's. It showed that FDLs for an 88-Hz target were not reduced when the components of an 88-Hz fringe were set to alternating phase, thereby doubling their pitch without affecting resolvability of the components. The implications for single- and dual-mechanism models of pitch will be discussed.