Abstract The literature suggests that up to 19% of umbilical cord blood samples are invalid. Accordingly, it has been proposed that blood should be universally collected from both vessels. We prospectively collected paired arterial and venous blood to examine whether our centre, where staff were experienced in single vessel collection, was more accurate. Of 289 paired samples, 53 (18.3%) were considered invalid. Despite this significant error rate, we propose that routinely, only arterial sampling is needed and that an additional venous sample need only be taken to validate samples in cases of pH < 7.15, a difficult delivery or a non-vigorous baby.