The 18 O enrichment (Δ 18 O) of cellulose (Δ 18 O Cel ) is recognized as a unique archive of past climate and plant function. However, there is still uncertainty regarding the proportion of oxygen in cellulose ( p ex ) that exchanges post‐photosynthetically with medium water of cellulose synthesis. Particularly, recent research with C 3 grasses demonstrated that the Δ 18 O of leaf sucrose (Δ 18 O Suc , the parent substrate for cellulose synthesis) can be much higher than predicted from daytime Δ 18 O of leaf water (Δ 18 O LW ), which could alter conclusions on photosynthetic versus post‐photosynthetic effects on Δ 18 O Cel via p ex . Here, we assessed p ex in leaves of perennial ryegrass ( Lolium perenne ) grown at different atmospheric relative humidity (RH) and CO 2 levels, by determinations of Δ 18 O Cel in leaves, Δ 18 O LGDZW (the Δ 18 O of water in the leaf growth‐and‐differentiation zone) and both Δ 18 O Suc and Δ 18 O LW (adjusted for ε bio , the biosynthetic fractionation between water and carbohydrates) as alternative proxies for the substrate for cellulose synthesis. Δ 18 O LGDZW was always close to irrigation water, and p ex was similar (0.53 ± 0.02 SE) across environments when determinations were based on Δ 18 O Suc . Conversely, p ex was erroneously and variably underestimated (range 0.02–0.44) when based on Δ 18 O LW . The photosynthetic signal fraction in Δ 18 O Cel is much more constant than hitherto assumed, encouraging leaf physiological reconstructions.