A novel pterin intermediate, in addition to the expected 4a-hydroxytetrahydrobiopterin (4a-OH-BH4) and quinonoid dihydrobiopterin, was generated during catalytic turnover of tyrosine hydroxylase (TH) with tetrahydrobiopterin as the cofactor. Based on chromatographic, spectroscopic and stability properties its structure is proposed to be similar to the product formed by the non-enzymic conversion of synthetic 4a-OH-BH4 [Bailey, Rebrin, Boerth and Ayling (1995) J. Am. Chem. Soc. 117, 10203-10211]. This compound was tentatively described as a 4a-adduct of a side-chain hydroxy group, i.e. the O2', 4a-cyclic-tetrahydrobiopterin (4a-Cyc-BH4). The intermediate generated in the TH reaction has a UV spectrum which is similar to that of 4a-OH-BH4, but elutes with a longer retention time (tR = 1.69 min compared with 1.06 min) on reversed-phase chromatography. Its conversion into quinonoid dihydrobiopterin is catalysed by pterin-4a-carbinolamine dehydratase (EC 4.2.1.96), although 4a-OH-BH4 is the preferred substrate for that enzyme. A precursor-product relationship was demonstrated between 4a-OH-BH4 and the putative 4a-Cyc-BH4 intermediate. The apparent stability of this compound is dependent on pH as well as on the nature of the buffer ions. At pH 8.0 a large amount was generated in Hepes and Tris, but little in phosphate buffer. At pH 7.0 in Hepes (standard assay conditions) and Tris buffer the putative 4a-Cyc-BH4, but no 4a-OH-BH4, was observed. None of the intermediates was observed at pH 6.0. The accumulation of these intermediates in the absence of dehydratase has important implications for the assay of TH and phenylalanine hydroxylase activities, and is also compatible with a possible physiological role of the dehydratase in the synthesis of catecholamines in vivo.