Studies by e.p.r. (electron-paramagnetic-resonance) spectroscopy and by stopped-flow spectrophotometry on turkey liver xanthine dehydrogenase revealed strong similarities to as well as important differences from the Veillonella alcalescens xanthine dehydrogenase and milk xanthine oxidase. The turkey enzyme is contaminated by up to three non-functional forms, giving molybdenum e.p.r. signals designated Resting I, Resting II and Slow. Slow and to a lesser extent Resting I signals are like those from the Veillonella enzyme, whereas Resting II is very like a resting signal described by K. V. Rajagopolan, P. Handler, G. Palmer & H. Beinert (1968) (J. Biol. Chem. 243, 3784-3796) for aldehyde oxidase. Another non-functional form that gives the Inhibited signal is produced on treatment of the enzyme with formaldehyde. Stopped-flow measurements at 450 nm show that, as for the milk enzyme, reduction by xanthine is rate-limiting in enzyme turnover. The active enzyme gives rise to Very Rapid and Rapid molybdenum(V) e.p.r. signals, as well as to an FADH signal. That these signals are almost indistinguishable from those of the milk enzyme, confirms the similarities between the active sites. There are two types of iron-sulphur centres that give signals like those in the milk enzyme, though with slightly different parameters. Quantitative reduction titration of the functional enzyme with xanthine revealed two important differences between the turkey and the milk enzymes. First, the turkey enzyme FADH/FADH2 system has a redox potential sufficiently low that xanthine is incapable of reducing the flavin completely. This finding presumably explains the very low oxidase activity. Secondly, whereas the Fe/S II chromophore in the milk enzyme has a relatively high redox potential, for the turkey enzyme the value of this potential is lower and similar to that of its Fe/S I chromophore.