Three recently reported chemiluminescence methods (based on reactions with alkaline luminol and hexacyanoferrate(III); acidic cerium(IV) and rhodamine B; and acidic permanganate with polyphosphates) for the determination of synephrine were re-evaluated in terms of their selectivity towards this analyte in comparison to other phenolic compounds. A fourth reagent system, acidic soluble manganese(IV) and formaldehyde, was also examined. Each set of reagents was sensitive towards synephrine (limits of detection were 3 x 10(-9), 5 x 10(-8), 1 x 10(-8) and 1 x 10(-8) mol/L, respectively) but also responded with numerous other phenolic compounds, including some that are present in citrus fruit extracts, dietary supplements and/or biological fluids. It is therefore recommended that the determination of synephrine in these matrices should incorporate physical separation of sample components (e.g. chromatography or electrophoresis). In more general terms, this study illustrates that accurate percentage recoveries for an analyte in spiked samples (without validation against another analytical method) are insufficient to confirm the analytical utility of new flow-injection analysis (FIA) procedures.