Femtosecond transient absorption spectroscopy has been applied to the isolated carotenoid β-carotene under a large variety of experimental conditions regarding solvent, temperature, excitation wavelength, and intensity to study the excited state relaxation dynamics in order to elucidate the origin of the so-called "dark S* state", which has been discussed very controversially in the literature. The results are analyzed in terms of lifetime density maps, and various kinetic models are tested on the data. The sample purification was found to be critical. The appearance of a component with a lifetime longer than that of the relaxed S(1) state (i.e., τ > 10 ps), which has been associated previously with the S* (or S(‡)) state is due to the presence of an impurity. For pure samples, four lifetimes are typically observed (all ≤10 ps at room temperature). Consideration of the large body of experimental data leads us to exclude relaxation schemes implying a separate "dark S* state" in β-carotene formed in parallel to the normal S(2) → S(1) relaxation scheme. Vibrational cooling in the S(1) state can explain fully all the features of the transient spectra on the picosecond time scale within a S(2) → S(1v) → S(1v') → S(1) → S(0) relaxation scheme without invoking any additional electronic or distinctly different conformational states. Thus, we exclude assignments of the previously reported "S* state" signals in β-carotene (i) to require the postulate of a separate electronic state, (ii) to require the postulate of a large conformational change and/or a partial cis configuration formed in the relaxation pathway, or (iii) to require a vibrationally excited ground state (GS) species. High intensity excitation leads in part to a two-photon excitation to the S(2N) state which upon relaxation gives rise to a different vibrational excitation pattern in the initially created hot S(1) state(s). The spectral changes in the S(1v) state observed upon both short wave excitation as well as high intensity excitation can be explained well by such a modified vibrational excitation pattern. In contrast, the variations in the difference spectra of the partially (S(1v')) and fully vibrationally relaxed S(1) states (S(1)) are minor. The data do not provide any evidence that would require one to postulate the existence of a separate "S* state".