In other words, [b]HOW[/b] the MP3s were originated, there is always the possibility of a couple of faults:
#1 The tracks in question have been downloaded from an online site and the downloads were faulty - how this would affect a CDA track I don't honestly know because downloads are always reported as faulty and usually deleted in preference for a faultless download.
#2 The tracks have been ripped by a program that didn't stop recording at the end of the track - it is possible to record to MP3 manually, whereby the recording won't stop until halted by the user doing the recording.
When an MP3 track is played with a utility such as GOMPlayer, or RealPlayer, or VLAN, or WinAMP, or Windows Media Player, or whatever... there is a capability built into the player, that when the last digital byte of music data is reached (playing volume falls below a preset decibel level), the Player automatically sequences forward to the next track in the playlist. The decibel level that prompts this function is built into the metadata of the MP3 file itself, and has long been a problem with MP3 recordings of passages of classical music which contain long periods of little or no volume within the musical track - in these cases the db level is set to almost nothing so that the 'playing' will continue, without jumping on to the next track.
This however will not happen with an AUDIO CD unless the track image has finished.
So - I again suggest you download and install Audacity:
http://audacity.sourceforge.net/download/
Load one of your problem tracks into Audacity and scroll to the end of the track. You'll be able to see if the music ends long before the track actually ends. If this is the case, you can chop the silent passage of the track, saving the trimmed version back to hard drive as an alternate MP3. :)