Quantitative estimates of gibberellin A9 in Norway spruce extracts obtained by gas chromatography-mass spectrometry, radioimmunoassay (RIA_ and bioassay were compared after successive purifications of the extracts. The extracts were assayed in several dilutions with and without the addition of standard gibberellin A9, thus showing the effect of extract components on the response of the assays. Radioimmunoassay produced estimates comparable to gas chromatography-mass spectrometry after one purification step by high-performance liquid chromatography. Extracts purified by polyvinylpyrrolidone-column chromatography and solvent partitioning but not high-performance liquid chromatography resulted in inaccurate RIA estimates. The performance of the RIA could be monitored by logit-log transformations of the standard curve and extract dilution curve and by calculating the slope of the standard addition curve. It was, however, not possible to correct for the interference caused by extract components by the standard addition procedure. Quantifications by Tan-ginbozu dwarf-rice bioassay were accurate, but a large and unpredictable variation makes it unsuitable for quantitative determinations.