Human apolipoprotein (apo) B mRNA is edited in a tissue specific reaction, to convert glutamine codon 2153 (CAA) to a stop translation codon. The RNA editing product templates and hybridises as uridine, but the chemical nature of this reaction and the physical identity of the product are unknown. After editing in vitro of [32P] labelled RNA, we are able to demonstrate the production of uridine from cytidine; [alpha 32P] cytidine triphosphate incorporated into RNA gave rise to [32P] uridine monophosphate after editing in vitro, hydrolysis with nuclease P1 and thin layer chromatography using two separation systems. By cleaving the RNA into ribonuclease T1 fragments, we show that uridine is produced only at the authentic editing site and is produced in quantities that parallel an independent primer extension assay for editing. We conclude that apo B mRNA editing specifically creates a uridine from a cytidine. These observations are inconsistent with the incorporation of a uridine nucleotide by any polymerase, which would replace the alpha-phosphate and so rule out a model of endonucleolytic excision and repair as the mechanism for the production of uridine. Although transamination and transglycosylation remain to be formally excluded as reaction mechanisms our results argue strongly in favour of the apo B mRNA editing enzyme as a site-specific cytidine deaminase.