The selective removal of water from mixtures with methanol, ethanol, and 1-propanol is an important task in the processing industries. With the aid of configurational-bias Monte Carlo simulations of unary and mixture adsorption, we establish the potential of CuBTC for this separation task. For operations close to pore saturation conditions, the adsorption is selective to water that has a significantly higher saturation capacity compared to that of 1-alcohols. The water-selective separation relies on subtle entropy effects that manifest near pore saturation conditions. A further distinguishing feature is that mixture adsorption is determined to be strongly nonideal, and the activity coefficients of the constituent components deviate strongly from unity as pore saturation is approached. The predictions of the ideal adsorbed solution theory (IAST), though qualitatively correct, do not predict the component loadings for mixture adsorption with adequate accuracy. Consequently, the activity coefficients, after appropriate parametrization, have been incorporated into the real adsorbed solution theory (RAST). Transient breakthrough simulations, using the RAST model as a basis, demonstrate the capability of CuBTC for selective adsorption of water in fixed-bed adsorption devices operating under ambient conditions.