Errors in the FITS file frequency definition may affect the u-v coordinates, and so may affect the RA/DEC scales on any images that might be produced from faulty visibility datasets.
One fairly innocuous AIPS frequency bug is that the bandwidth of `channel 0' datasets can be wrong. Leaving this uncorrected will invalidate error estimates generated by Miriad tasks - which is not a major problem if you ignore these anyway!
Another, more serious, problem is that two locations in the FITS file give the ``reference frequency'' - the header and the antenna table. In principle these should agree. However there are instances when they will differ. Task fits will issue a warning when this happens, and will use the antenna table reference frequency as the true reference frequency. In some cases this will be the correct frequency. In others it will not.
There are two approaches to fixing the errors in the frequency definitions - either correct them in Miriad (using puthd and uvputhd) after reading the FITS file, or in AIPS (using PUTHEAD and TABED) before writing the FITS file. Which you will use will depend on your familiarity with the AIPS and Miriad visibility dataset structures and the complexity of your dataset (how many frequencies, etc), and where the bug occurred (are the u-v coordinates correct?). It is best to seek help, both to fix your data, and to aid tracking down the AIPS bugs.