The chemical shifts of the geminal proton resonances of H(2)-26 (δ(a) and δ(b)) are a widely used predictor of C-25 stereochemistry in furostanol steroidal saponins, being in general more resolved in 25S than 25R compounds. Unexpectedly, we found that application of this empirical rule in different solvents led to conflicting assignments of stereochemistry. An experimental survey revealed that, while the chemical shifts of H(2)-26 exhibit a dependence on C-25 configuration, it is less pronounced in methanol-d(4) than pyridine-d(5) solvent, and thus the general rule derived for pyridine-d(5) fails when NMR spectra are acquired in methanol-d(4). We propose a modified empirical method for the direct assignment of C-25 stereochemistry in furostanol saponins in methanol-d(4) (Δ(ab)=0.45-0.48ppm for 25S; Δ(ab)=0.33-0.35ppm for 25R), and provide several detailed examples. In addition, the absolute configuration of compound 8, a steroidal saponin isolated in previous work from Ruscus colchicus, is corrected from 25R to 25S stereochemistry.