AHA! It was a 3dTcat error!
3dinfo showed 133 time points for each of the 8 EPI's, and should have shown 133x8=1064 for the concatenated dataset. But instead, this contained 1064+133=1197 timepoints. I must have accidentally typed in one of the EPI's twice in my 3dTcat command line, resulting in concatenation of 9 EPI's, one of which was redundant.
Many thanks to Rick and Gang for your help in my detective work.
-Whitney