Fluorescence resonance energy transfer (FRET) is a widely utilized optical technique for measuring small distances of 1-10 nm in live cells. In recent years, its application has been greatly popularized by the discovery of green fluorescent protein (GFP) and many improved variants which make good donor-acceptor fluorophore pairs. GFP-based proteins are structurally stable, relatively inert, and can be reliably attached to points of interest. The combination of easy access to the GFP-based FRET technique and its obvious usefulness in many applications can lead to complacency. Potential problems such as light contaminants, e.g., bleed-through and cross-talk, and inconsistent donor and acceptor concentrations are easily overlooked and can lead to errors in FRET calculation and data interpretation. In this article, we outline possible pitfalls of GFP-based FRET and approaches that address these issues, including a "Spectra FRET" technique that can be easily applied to live cell studies.