This article describes a method for the enzymatic detection of low-abundant metabolic intermediates in plant extracts via NAD(P)H fluorescence using a microtiter plate reader. The detection of changes in NAD(P)H fluorescence (excitation 340 nm, emission 465 nm) exhibits a high signal-to-noise ratio and is as sensitive (> or = 20 pmol per well) as absorbance measurements with dual-wavelength photometers. Since up to 96 reactions can be initiated, monitored, and evaluated simultaneously, this method might be suitable for high-throughput screening programs on metabolite profiles. However, in contrast to absorbance measurements, fluorescence detection of NAD(P)H yields relative data, which can be impaired by the quench characteristics and the basic fluorescence of the extracts. Hence, extensive calibration is required to gain reproducible results. Calibration of the assay system was performed using leaf or root material (equivalent to 2-35 mg of fresh weight per well) extracted with perchloric acid, chloroform/water/methanol, or hot ethanol. Extraction with perchloric acid was found to be superior for metabolite quantification. Examples of the kinetics of individual metabolite determinations are presented and the contents of 3-phosphoglycerate, hexose phosphates, triose phosphates, pyruvate, and phosphoenolpyruvate in illuminated and darkened spinach leaves as well as leaf rosettes of Arabidopsis thaliana and leaf segments of the inducible crassulacean acid metabolism plant Mesembryanthemum crystallinum were measured via NAD(P)H fluorescence and, where possible, compared to reported data determined with dual-wavelength photometers.