The diethyl ether-extracted organic compounds of 155 samples of unifloral grade New Zealand kamahi and honeydew honeys, and New Zealand and Norwegian erica honeys, together with a series of active and inactive manuka honeys were analysed using combined gas chromatography/mass spectrometry. It was found that Kamahi honey is characterized by the presence of 2,6-dimethylocta-3,7-diene-2,6-diol, meliracemoic acid, and kamahines A-C and these compounds were typically present at average levels of 31, 14, and 73 mg/kg of honey, respectively. 2,6-Dimethylocta-3,7-diene-2,6-diol was isolated and the structure of this compound was defined using one- and two-dimensional NMR analyses. The only recognizably distinct peak present in the honeydew honey profile was indole acetic acid. In this honey, a relatively low to moderate level of indole acetic acid, ranging from 0.9 to 9.1 mg/kg honey was detected. In the New Zealand erica honey samples, ericinic acid, isoericinic acid isomers (average levels 363 and 34 mg/kg respectively), trans,cis and trans,trans-abscisic acid isomers (average levels 302 and 224 mg/kg respectively) and benzoic acid (average level 6950 mg/kg) were identified as floral marker compounds. Ericinic acid was isolated and the structure of this acid was defined using one-and two-dimensional NMR analyses. Low levels of ericinic and isoericinic acids (average levels of 1.1 and 0.32 mg/kg respectively) were detected in the Norwegian erica-rich honeys. The results presented here indicate that ericinic and isoericinic acids are likely to be universally present in erica honeys at levels which may range from as low as 1 mg/kg or less, as found in some Norwegian samples, to more than 100 mg/kg in some New Zealand samples. Two groups, namely a fingerprint pattern which characterized active manuka honeys, and a fingerprint pattern that characterized inactive manuka honeys were identified. Some substances contributing to the GCMS profile were found as marker compounds for the presence of unidentified substances responsible for the UMF activity. A statistically significant correlation was found between a small set of marker compounds (i.e. phenylacetic acid, 2-methoxyacetophenone, 2-methoxybenzoic, phenyllactic, octanedioic, cis-cinnamic, trans-cinnamic, nonanedioic, 4-methoxyphenyllactic and decanedioic acids and methyl syringate) and UMF activity of manuka honey. The best-fit marker compound regression equation (R = 0.92) was obtained for a set of pooled 30 moderate to high activity (UMF gt 14.1) samples. It was shown that the marker compound regression equation is capable of predicting the approximate UMF activity in both active and inactive manuka and kanuka honey samples. The leaf oil profiles of manuka (L. scoparium) plants that yielded active and inactive manuka honeys were characterized using an adaption of the micro-scale extraction and GC/FID or GC/MS, technique developed by Brophy et al. (1989). Six major groups of volatile (steam distillable) compounds (monoterpenes, sesquiterpene hydrocarbons, oxygenated sesquiterpenes [excluding eudesmols], eudesmols, triketones, and nor-triketones) and 3 groups of non-volatile or semi-volatile compounds (flavonoids, grandiflorone and nor-grandiflorone) were recognized in the leaf oil components. The active manuka honeys do not appear to be derived uniquely, or predominantly, from a single leaf oil chemotype.