You probably mean you are using a twice-refocused spin echo sequence as in this paper by Reese et al. 2003:
http://www.ncbi.nlm.nih.gov/pubmed/12509...

This does a good job of reducing eddy current distortion.

There are still distortions from other sources such as geometric distortion from magnetic susceptibility differences
http://www.ncbi.nlm.nih.gov/pubmed/76749...
and there is likely to be motion during the acquisition which should be corrected.