Molecular simulation as a complementing tool to experimental measurements has grown to be an instrument with predictive power for ionic liquids (ILs) derivatives. The combination of molecule-specific atomic charges generated with ab initio calculations and bonded and vdW parameters extracted from pre-fitted transferable force fields serves as the common practice in this field. Charge scaling is often applied to account for charge transfer and polarization effects and the scaling factor 0.8 has been concluded as a near-optimal solution in many scientific reports based on detailed analyses of various structural and thermodynamic observables. However, for ILs from the pyrrolidinium family investigated in the current work, we show that the charge scaling treatment cannot really minimize the experiment-simulation deviations to an acceptable level, which calls for further parameter adjustment. Following the modelling protocol summarized in our previous works, we tune the vdW radius of ILs-forming ions with a 298 K density scan and secure a parameter set that satisfactorily reproduces not only the room-temperature properties but also the mass densities in a large temperature range from 288 K to 348 K. The slopes of the vdW-scaling profiles are between −1.9 g/cm3 and −1.6 g/cm3, which are similar to quinuclidinium ILs investigated in our previous work. This phenomenon suggests the similarity of responding behaviors of the two ILs family during vdW scaling, and this range of response strengths could possibly be representative for ILs vdW scaling. Aside from bulk solvents, we further consider the interaction of ILs with a series of external organic agents. Following the previously accumulated experience, a slightly modified vdW scaling factor is picked as an option balancing different interactions within the heterogeneous solution. Nonequilibrium alchemical free energy calculations are conducted to compute solvation free energies in ILs and also the distribution (partition) behaviors of solutes between aqueous and ILs phases. The computed physiochemical properties involving pyrrolidinium ILs agree well with the experimental reference. Another feature of the selected pyrrolidinium ILs is the systematic variation of the length of the alkyl substituting chain, which enables a direct investigation of the chain-length dependence of the prediction accuracies for both bulk density and solvation/partition thermodynamics. Interestingly, with the chain-length variation, the error of bulk density exhibits a monotonic response, while a fluctuating (non-monotonic) behavior is observed for prediction errors of solvation and partition thermodynamics, which is related to the diverse features of solute-solvent interactions. Therefore, the observed behaviors of the properties of bulk solvents cannot be generalized to the solvation/partition behaviors of the same solvent, and accurate reproductions of bulk behaviors do not guarantee accurate calculations of solvation and partition thermodynamics.