Carlyon and Shackleton [J. Acoust. Soc. Am. 95, 3541-3554 (1994)] suggested that fundamental-frequency (F0) discrimination performance between resolved and unresolved harmonics is limited by an internal "translation" noise between the outputs of two distinct F0 encoding mechanisms, in addition to the encoding noise associated with each mechanism. To test this hypothesis further, F0 difference limens (DLF0s) were measured in six normal-hearing listeners using sequentially presented groups of harmonics. The two groups of harmonics presented on each trial were bandpass filtered into the same or different spectral regions, in such a way that both groups contained mainly resolved harmonics, both groups contained only unresolved harmonics, or one group contained mainly resolved and the other only unresolved harmonics. Three spectral regions (low: 600-1150 Hz, mid: 1400-2500 Hz, or high: 3000-5250 Hz) and two nominal F0s (100 and 200 Hz) were used. The DLF0s measured in across-region conditions were well accounted for by a model assuming only two sources of internal noise: the encoding noise estimated on the basis of the within-region results plus a constant noise associated with F0 comparisons across different spectral regions, independent of resolvability. No evidence for an across-pitch-mechanism translation noise was found. A reexamination of previous evidence for the existence of such noise suggests that the present negative outcome is unlikely to be explained by insufficient measurement sensitivity or an unusually large across-region comparison noise in the present study. While the results do not rule out the possibility of two separate pitch mechanisms, they indicate that the F0s of sequentially presented resolved and unresolved harmonics can be compared internally at no or negligible extra cost.