Time domain correlations of common features among relative paleointensity records from sedimentary cores are invaluable to paleomagnetism and paleoclimatology. Sediments with high accumulation rates might now provide millennial scale correlations of temporal variations in the geomagnetic dipole moment. Errors in the ages of paleomagnetic data samples, however, can make such correlations difficult and unreliable. We use spectral methods to assess the level of coherence expected among individual and stacked high-resolution simulated paleointensity records for the time interval 0--75 ka. Correlations between individual paleointensity records are systematically degraded with decreased sedimentation rate and increased magnitude of age errors. We find that with optimistic age errors and interpolation of depth sampled data to evenly spaced time series, only short period signal in high-resolution relative paleointensity is corrupted. For currently available methods of establishing chronologies, we estimate the minimum characteristic timescale of correlative features between pairs of regional stacked records at about 4.5 kyr. From an analysis of NAPIS-75 and SAPIS data, it appears that the limit is inherent to the regional stacks and not a consequence of comparison of distant, independent data sets. A detailed comparison of the NAPIS-75 and SAPIS stacks shows that this limit is likely larger, perhaps 6 kyr. At long periods the two regional stacks are more poorly correlated than those from our simulations, suggesting somewhat larger age errors in the individual paleointensity records.