The electronic and geometrical structures of the ground and exited states of Fe-n, Fe-n(-), and Fe-n(+) are computed by density functional theory. Because the assignment of the ground states of Fe-3, Fe-3(-), Fe-4, and Fe-4(-) is controversial, these systems are studied using several different functionals. It appears that the LSDA and B3LYP methods do not work well for iron clusters and should be avoided. The number of unpaired electrons in the neutral ground states is 6(Fe-2), 10(Fe-4) 14(Fe-4), 16(Fe-5), and 20(Fe-6). The number of unpaired electrons in the ground states of the anions and cations differ by one from the corresponding neutral, except for Fe-4(+), which has three fewer unpaired electrons than Fe-4. The computed DFT adiabatic electron affinities and ionization potentials of the neutral clusters are in good agreement with experiment. Fragmentation energies are in qualitative agreement with experiment, where the error is about 1 eV for the dissociation energy of the iron dimer. The natural bond analysis allows one to qualitatively understand the nature of high local magnetic moments at iron sites and their evolution from Fe-2 to Fe-6.