The prediction of the existence and stability of (meta)-stable phases in a chemical system is realized via a two-step process: identification of structure candidates through global exploration of the classical empirical energy landscape, followed by a local optimization of the candidates on ab-initio level employing a heuristic algorithm. From the computed energy/volume curves, one can then calculate the thermodynamically stable phase at a given pressure and the transition pressures among the phases. In order to gain insight into the kinetic stability of the structure candidates, one computes estimates of the energy and enthalpy barriers around the structures with the so-called threshold algorithm, yielding a tree graph representation of the chemical system. In this work we perform a theoretical and experimental study of the LiI energy landscape. We determine the structure candidates, construct the tree graph representation and compute the abinitio energy/volume curves for the hypothetical structures. We find that the thermodynamically preferred modifications at standard pressure should exhibit the rock salt and the wurtzite structure, respectively. In order to validate our predictions by experiments, we have employed the newly developed 'Low-Temperature-Atomic Beam Deposition' (LT-ABD) technique, which allows to disperse the components of the desired product at an atomic level and in an appropriate ratio. After depositing LiI at T = 77 K, the first crystallization occurs at T approximate to 173 K in the wurtzite-type structure followed by a transition to the more stable rock salt-type structure at T approximate to 273 K. At room temperature only the cubic phase remains.