Both FRET and BiFC are effective methods in detecting protein-protein interactions (PPI) in vivo and in vitro.  FRET has been in use for a longer time than BiFC, so it would be reasonable to assume that it has undergone more refinement than BiFC. BiFC has the disadvantage that the two halves of the split flourophore protein (e.g., GFP) might autonomously and spontaneously reassemble, which will increase the false positive rate in the PPI detection experiment.  Have a look at the article:

http://www.plantphysiol.org/content/171/2/727.long

Of note is the fact that FRET and BiFC have recently been combined as a PPI detection method, and is perhaps better than either one alone.

FRET is reversible and is more powerful in determining spatio-temporal dynamics of a protein-protein interaction. BiFC probably has this disadvantage and once a transient complex between the fused protein to the two halves of BiFC is formed, the reconstituted fluorescent protein remains stable. Another major difference is FRET can in real time, but BiFC reconstitution would take time for the chromophore to mature. Hence, as I said earlier FRET is more effective in determining temporal changes in PPI.