Evidence is presented that intermediates of the oxidative pentose phosphate pathway (OPPP) are channeled from one pathway enzyme to the next. CO, produced from [l-'4C]glucose in the presence of unlabelled pathway intermediates contained much more radioactivity than predicted by a model in which pathway-produced intermediates are in equilibrium with identical molecules in the bulk phase. This was the case whether glucose 6-phosphate (Glc6P), 6-phosphogluconolactone, or 6-phosphogluconate was added. Assumptions involved in calculating the amount of I4CO, predicted for free mixing of I4C-labelled and unlabelled intermediates are discussed, together with the following results. (a) 14C0, production by pea nodules in the presence of 3 mM 6-phosphogluconate was higher than in its absence. (b) Apparent channeling of intermediates was much higher for purified yeast enzymes than for yeast extract. (c) 6-Phosphogluconate and 6-phosphogluconolactone were channeled between yeast Glc6P dehydrogenase and 6-phosphogluconate dehydrogenase despite the absence of 6-phosphogluconolactonase in the purified yeast enzyme mixture. (d) When purified yeast hexokinase was physically separated from Glc6P dehydrogenase and 6-phosphogluconate dehydrogenase by a dialysis membrane, there was no apparent channeling. (e) Poly(ethy1ene glycol), high salt and detergents had little effect on apparent channeling of OPPP intermediates, which is consistent with a stable complex of enzymes. On the other hand, density gradient centrifugation experiments suggested a more transient interaction between the enzymes. Taken together, the results support channeling of OPPP pathway intermediates.