Most ionic liquids are known to be hygroscopic to varying degrees, and that can be detrimental or useful depending upon the application in question. Water can accumulate slowly over hours or days to saturation levels corresponding to the humidity level. When designing or deploying a new ionic liquid it is important to be able to estimate its maximum moisture absorbing ability at the temperature and pressure of its operation. With this goal in mind we have carried out computational studies on three ionic liquid systems based on [BF₄]⁻, [PF₆]⁻, and [Tf₂N]⁻ anions and 1-alkyl-3-methyl-imidazolium ([C_nmim]⁺) cations within an implicit solvent formalism. For highly hygroscopic systems like [C_nmim][BF₄] we find that non-iterative calculations with single water molecules can lead to significant underestimation of the maximum moisture content, while iterative calculations can result in miscibility behavior qualitatively different from experimental observations. On the other hand, the inclusion of small hydrogen-bonded water-clusters up to an appropriately chosen size is shown to yield better quantitative agreements with experimentally observed water uptake. Additionally, such calculations appear consistent with a number of thermodynamically interesting phase behaviors, including limited-solubility to full-miscibility transitions as a function of temperature and as a function of the alkyl chain length of the imidazolium cation. For hydrophobic systems like [C_nmim][PF₆] and [C_nmim][Tf₂N] the computed solubility (for each n) is found to have a smooth convergence behavior as a function of the largest cluster-size considered with the results for the larger clusters being close to that obtained by iterative calculations with single water molecules.