A method for rapid identification of bacterial cytokinins has been developed in which cultures are fed [3H]adenine, the cytokinins (including 3H-labeled cytokinins) are isolated by immunoaffinity chromatography, and analyzed by HPLC with on-line scintillation counting. Analysis of Agrobacterium tumefaciens strains showed that some produced primarily trans-zeatin, whereas others produced primarily trans-zeatin riboside. Pseudomonas syringae pv savastanoi produced mixtures of transzeatin, dihydrozeatin, 1"-methyl-trans-zeatin riboside, and other unknown cytokinin-like substances. Corynebacterium fascians, produced cis-zeatin, isopentenyladenine and isopentenyladenosine. The technique is designed for qualitative rather than quantitative studies and allows ready identification of bacterial cyto- 'Abbreviations: RIA, radioimmunoassay; "MeZR, 1 "-methyl trans-zeatin riboside; cZ, cis-zeatin; cZR, cis-zeatin riboside; DZ, dihydrozeatin; DZR, dihydrozeatin riboside; iP, isopentenyladenine; iPA, isopentenyladenosine; Z, trans-zeatin; ZR, trans-zeatin riboside.Unfortunately, direct analysis of phytohormones in unfractionated plant extracts by RIA or ELISA can lead to erroneous results. In some cases, crude extracts may contain compounds which cause nonspecific interference; their presence leads to generation of false positives. For example, Cohen et al. (6) showed that ELISA ofIAA in even semipurified plant samples led to high estimates of the IAA content. As successively more pure samples were analyzed, the apparent IAA values approached those determined by rigorous mass spectrometric analysis. Nonspecific interference is, however, not the only problem. When several cross-reacting phytohormone species are present, a composite result will be obtained. A case in point is the analysis of bacterially produced cytokinins. The primary cytokinin produced by Agrobacterium tumefaciens strain C58 has been shown to be zeatin (9). Small amounts of iP are also present, but not in sufficient quantity to interfere with zeatin-specific RIA of supernatants (20). Zeatin determination by ELISA or RIA is therefore likely to give reasonably accurate results. The same is not true for Pseudomonas syringae pv savastanoi strains. Some strains secrete Z, ZR, and 1 "MeZR and, late in stationary phase, these are present in the medium in approximately equal quantities (14). Since the relative cross reactivities of Z, ZR and 1 "MeZR with anti-ZR antibody are approximately 10:50:1 (unpublished data), assays ofunfractionated supernatants will not give meaningful data. For most plant samples, where several cytokinins are usually present, the same considerations apply.The problem can be alleviated by combining immunoaffinity chromatography, to purify the cytokinins from nonspecific interfering substances (13), with HPLC to separate the individual components. RIA or ELISA can then be used to detect each cytokinin in the appropriate HPLC fraction (7,12