Despite their fundamental biological relevance, structure-property relationships in N-glycans are fundamentally lacking, and their highly multidimensional compositional and conformational phase spaces remain largely unexplored. The torsional flexibility of the glycosidic linkages and the ring dynamics result in wide, rugged free-energy landscapes that are difficult to sample in molecular dynamics simulations. We show that a novel enhanced-sampling scheme combining replica exchange with solute and collective-variable tempering, enabling transitions over all relevant energy barriers, delivers converged distributions of solvated N-glycan conformers. Several dimensionality-reduction algorithms are compared and employed to generate conformational free-energy maps in two dimensions. Together with an originally developed conformation-based nomenclature scheme that uniquely identifies glycan conformers, our modeling procedure is applied to reveal the effect of chemical substitutions on the conformational ensemble of selected high-mannose-type and complex glycans. Moreover, the structure-prediction capabilities of two commonly used glycan force fields are assessed via the theoretical prediction of experimentally available nuclear magnetic resonance J-coupling constants. The results especially confirm the key role of ω and ψ torsion angles in discriminating between different conformational states and suggest an intriguing correlation between the torsional and ring-puckering degrees of freedom that may be biologically relevant.