We have experimentally studied the fluorescence resonance energy transfer (FRET) between green fluorescent protein (GFP) molecules by inserting folded or intrinsically unstructured proteins between CyPet and Ypet. We discovered that most of the enhanced FRET signal previously reported for this pair was due to enhanced dimerization, so we engineered a monomerizing mutation into each. An insert containing a single fibronectin type III domain (3.7 nm end-to-end) gave a moderate FRET signal while a two-domain insert (7.0 nm) gave no FRET. We then tested unstructured proteins of various lengths, including the charged-plus-PQ domain of ZipA, the tail domain of alpha-adducin, and the C-terminal tail domain of FtsZ. The structures of these FRET constructs were also studied by electron microscopy and sedimentation. A 12 amino acid linker and the N-terminal 33 amino acids of the charged domain of the ZipA gave strong FRET signals. The C-terminal 33 amino acids of the PQ domain of the ZipA and several unstructured proteins with 66-68 amino acids gave moderate FRET signals. The 150 amino acid charged-plus-PQ construct gave a barely detectable FRET signal. FRET efficiency was calculated from the decreased donor emission to estimate the distance between donor and acceptor. The donor-acceptor distance varied for unstructured inserts of the same length, suggesting that they had variable stiffness (persistence length). We conclude that GFP-based FRET can be useful for studying intrinsically unstructured proteins, and we present a range of calibrated protein inserts to experimentally determine the distances that can be studied.