The kinase responsible for the phosphorylation of d-ribulose was purified 45.5-fold from a strain of Escherichia coli K-12 capable of growth on d-arabinose with no separation of d-ribulo- or l-fuculokinase activities. Throughout the purification, the ratios of activities remained essentially constant. A nonadditive effect of combining both substrates in an assay mixture; identical K(m) values for adenosine triphosphate with either l-fuculose or d-ribulose as substrate; and, the irreversible loss of activity on both substrates, after removal of magnesium ions from the enzyme preparation, suggest that the dual activity is due to the same enzyme. A fourfold greater affinity of the enzyme for l-fuculose than for d-ribulose, as well as a higher relative activity on l-fuculose, suggest that the natural substrate for this enzyme is l-fuculose. The product of the purified enzyme, with d-ribulose as substrate, was prepared. The ratio of total phosphorous to ribulose phosphate was 1.01:1, indicating that the product was ribulose monophosphate. The behavior of the kinase product in the cysteine-carbazole and orcinol reactions, as well as the results of periodate oxidation assays, provided evidence that it was not d-ribulose-5-phosphate. Reaction of this compound with a cell-free extract of E. coli possessing l-fuculose-l-phosphate aldolase activity resulted in the production of dihydroxyacetone phosphate and glycolaldehyde. The kinase product failed to reduce 2,3,5-triphenyltetrazolium and possessed a half-life of approximately 1.5 min in the presence of 1 n HCl at 100 C. These properties suggested that the phosphate group was attached to carbon atom 1 of d-ribulose.