The intrinsic fluorescence of nucleic acids is extremely weak compared to that of the fluorescent labels used to probe their structural and functional behavior. Thus, for technical reasons, the investigation of the intrinsic DNA fluorescence was limited for a long time. But with the improvement in spectroscopic techniques, the situation started to change around the turn of the century. During the past two decades, various factors modulating the static and dynamic properties of the DNA fluorescence have been determined; it was shown that, under certain conditions, quantum yields may be up 100 times higher than what was known so far. The ensemble of these studies opened up new paths for the development of label-free DNA fluorescence for biochemical applications. In parallel, these studies have shed new light on the primary processes leading to photoreactions that damage DNA when it absorbs UV radiation. We have been studying a variety of DNA systems, ranging from the monomeric nucleobases to double-stranded and four-stranded structures using fluorescence spectroscopy. The specificity of our work resides in the quantitative association of the steady-state fluorescence spectra with time-resolved data recorded from the femtosecond to the nanosecond timescales, made possible by the development of specific methodologies. Among others, our fluorescence studies provide information on the energy and the polarization of electronic transitions. These are valuable indicators for the evolution of electronic excitations in complex systems, where the electronic coupling between chromophores plays a key role. Highlighting collective effects that originate from electronic interactions in DNA multimers is the objective of the present Account. In contrast to the monomeric chromophores, whose fluorescence decays within a few picoseconds, that of DNA multimers persists on the nanosecond timescale. Even if long-lived states represent only a small fraction of electronic excitations, they may be crucial to the DNA photoreactivity because the probability to reach reactive conformations increases over time, owing to the incessant structural dynamics of nucleic acids. Our femtosecond studies have revealed that an ultrafast excitation energy transfer takes place among the nucleobases within duplexes and G-quadruplexes. Such an ultrafast process is possible when collective states are populated directly upon photon absorption. At much longer times, we discovered an unexpected long-lived high-energy emission stemming from what was coined "HELM excitons". These collective states, whose emission increases with the duplex size, could be responsible for the delayed fluorescence of ππ* states observed for genomic DNA. Most studies dealing with excited-state relaxation in DNA were carried out with excitation in the absorption band peaking at around 260 nm. We went beyond this and also performed the first time-resolved study with excitation in the UVA spectral range, where a very weak absorption tail is present. The resulting fluorescence decays are much slower and the fluorescence quantum yields are much higher than for UVC excitation. We showed that the base pairing of DNA strands enhances the UVA fluorescence and, in parallel, increases the photoreactivity because it modifies the nature of the involved collective excited states.