This paper focuses on the possibility to inject large volumes (up to 10 L) in ultra-high performancesupercritical fluid chromatography (UHPSFC) under generic gradient conditions. Several injection andmethod parameters have been individually evaluated (i.e. analyte concentration, injection volume, initialpercentage of co-solvent in the gradient, nature of the weak needle wash solvent, nature of the samplediluent, nature of the column and of the analyte). The most critical parameters were further investigatedusing in a multivariate approach. The overall results suggested that several aprotic solvents includingmethyl tert-butyl ether (MTBE), dichloromethane, acetonitrile or cyclopentyl methyl ether (CPME) werewell adapted for the injection of large volume in UHPSFC, while MeOH was generally the worst alternative.However, the nature of the stationary phase also had a strong impact and some of these diluents did notperform equally on each column. This was due to the existence of a competition in the adsorption of theanalyte and the diluent on the stationary phase. This observation introduced the idea that the samplediluent should not only be chosen according to the analyte but also to the column chemistry to limitthe interactions between the diluent and the ligands. Other important characteristics of the “ideal” SFCsample diluent were finally highlighted. Aprotic solvents with low viscosity are preferable to avoid strongsolvent effects and viscous fingering, respectively. In the end, the authors suggest that the choice of thesample diluent should be part of the method development, as a function of the analyte and the selectedstationary phase.