This project was the very beginning of research into the chemistry of eastern Australian banksia flowers. Using dynamic headspace sampling (DHS) analysis, differences in volatile components, consistent with detection of differences in odour, were detected among three different species and one commercial cultivar. Infraspecific variation was also observed between two known subspecies of Banksia ericifolia and between differently coloured forms of Banksia spinulosa var. collina. The cultivar, Banksia 'Giant Candles', was shown to have some of the chemical components of each of its supposed ancestors. The absence of known wound-response chemicals indicated that this DHS method was successful in leaving the inflorescences undamaged throughout the sampling procedure. The Likens-Nickerson modification of classical hydrodistillation methods was useful. The static headspace method (SHS) was easily automated and was shown to be chemically robust and sufficiently sensitive to detect volatile compounds from only a few flowers. The milder DHS method, which minimised mechanical and heat damage to the plant tissue, produced a different set of results. From the results of this project, a suite of volatile compounds has been proposed that may be useful in future behavioural studies to help determine whether animals are attracted to components of banksia odours. These candidates include some compounds that have been reported in animal secretions, wound-response chemicals that may be produced by the plant to aid its communication with other organisms, and a compound (suggested to be sulfanylmethyl acetate) not previously reported from natural sources. The mildest of the three analytical methods used, dynamic headspace sampling, was shown to be suitable for the potential chemotaxonomic evaluation of some members of the Banksia genus. / Doctor of Philosophy (PhD)