The calculation of spin-state splittings of transition-metal complexes is known to be very sensitive to the level of theory and the quality of the basis set. The performance of density functionals is in general understood, with older pure functionals overstabilizing low spin states, and Hartree-Fock and hybrid functionals doing the same for the high spin states. More recent functionals (OLYP/OPBE, TPSSh, SSB-D, M06-L) seem to improve on these, but there are still some systems that prove difficult. This Perspective describes some of the recent successes and pitfalls in the description of spin states and treats both methodology and applications such as metal-oxo complexes, exchange-enhanced reactivity, spin-state catalysis, and spin-crossover compounds.