In this work, we present a study of the excitation energies of adenine, cytosine, guanine, thymine, and the adenine-thymine (AT) and guanine-cytosine (GC) base pairs using long-range corrected (LC) density functional theory. We compare three recent LC functionals, BNL, CAM-B3LYP, and LC-PBE0, with B3LYP and coupled cluster results from the literature. We find that the best overall performance is for the BNL functional based on LDA. However, in order to achieve this good agreement, a smaller attenuation parameter is needed, which leads to nonoptimum performance for ground-state properties. B3LYP, on the other hand, severely underestimates the charge-transfer (CT) transitions in the base pairs. Surprisingly, we also find that the CAM-B3LYP functional also underestimates the CT excitation energy for the GC base pair but correctly describes the AT base pair. This illustrates the importance of retaining the full long-range exact exchange even at distances as short as that of the DNA base pairs. The worst overall performance is obtained with the LC-PBE0 functional, which overestimates the excitations for the individual bases as well as the base pairs. It is therefore crucial to strike a good balance between the amount of local and long-range exact exchange. Thus, this work highlights the difficulties in obtained LC functionals, which provides a good description of both ground- and excited-state properties.