To determine the application limits of green fluorescent protein (GFP) as a reporter gene or protein tag, we expressed GFP by itself and with fusion protein partners, and used three different imaging methods to identify GFP fluorescence. In conventional epifluorescence photomicroscopy, GFP expressed in cells could be distinguished as a bright green signal over a yellow-green autofluorescence background. In quantitative fluorescence microscopy, however, the GFP signal is contaminated by cellular autofluorescence. Improved separation of GFP signal from HeLa cell autofluorescence was achieved by the combination of confocal scanning laser microscopy using 488-nm excitation, a rapid cut-on dichroic mirror and a narrow-bandpass emission filter. Two-photon excitation of GFP fluorescence at the equivalent of approximately 390 nm provided better absorption than did 488-nm excitation. This resulted in increased signal/background but also generated a different autofluorescence pattern and appeared to increase GFP photobleaching. Fluorescence spectra similar to those of GFP alone were observed when GFP was expressed as a fusion protein either with glutathione-S-transferase (GST) or with glucokinase. Furthermore, purified GST.GFP fusion protein displayed an extinction coefficient and quantum yield consistent with values previously reported for GFP alone. In HeLa cells, the cytoplasmic GFP concentration must be greater than approximately 1 microM to allow quantifiable discrimination over autofluorescence. However, lower expression levels may be detectable if GFP is targeted to discrete subcellular compartments, such as the plasma membrane, organelles or nucleus.