High-spin lithium clusters, n+1Lin (n = 2−21), have been systematically studied by using density functional theory. Although these high-spin clusters have no bonding electron pairs, they are stable with respect to isolated atoms. A set of 42 density functional theory functionals were benchmarked against CCSD(T)/cc-pVQZ results for clusters from the dimer to the hexamer. For these clusters, the strong non-additivity on the binding energy is analyzed employing a many-body energy decomposition scheme, concluding that most of the binding energy is due to a balance between the three- and four-body contributions. After a quality parameter had been defined, the LC-BP86 functional was identified as the most promising one for the description of high-spin lithium clusters. We employ the dependence of the second energy difference on cluster size to predict the formation of a higher-stability cluster.