We have measured resonance Raman spectra, including absolute scattering cross sections, for I3- in ethanol at eleven excitation wavelengths from 266 to 416 nm. This spectral region spans two broad electronic transitions with dissociate to I2- + I in the lower energy band and I2- + I in the higher energy band. Recent femtosecond experiments by Ruhman and coworkers have shown that the photodissociation is complex, even though the precursor is only a triatomic. Figure 1 shows the resonance Raman spectrum obtained with excitation at 309 nm. We observe a long progression in the symmetric stretch (112 cm-1) on both the Stokes and anti-Stokes sides of the Rayleigh line. Only a weak progression is observed in the antisymmetric stretch (143 cm-1), the nominal reaction coordinate. Qualitatively, this indicates that the nuclear motions immediately following electronic excitation involve mainly a symmetric lengthening of both I-I bonds, with the asymmetry leading to product formation developing more slowly. In addition to the sharp Raman lines, we observe a broad underlying fluorescence that is not significantly Stokes-shifted with respect to the Raman lines. In fact, there is even weak anti-Stokes fluorescence. This indicates that electronic dephasing is competitive with nuclear motion away from the Franck-Condon region. Since this motion occurs in around 10-20 fs, the electronic dephasing time must also be on this order. To give quantitative strength to these qualitative interpretations, we have calculated the absorption spectra and Stokes and anti-Stokes resonance Raman excitation profiles using the time-dependent wave-packet formalism of Heller and co-workers for several models of the excited state surfaces. These calculations include thermal population in higher vibrational levels of the ground state. The first surface explored was a LEPS potential constrained to fit the absorption spectrum and to give the correct asymptotic I2- ground state surface. While this surface crudely predicts the observed intensity distributions, the agreement is not quantitative. The latter assumption is in clear contrast to the experiment observation of unrelaxed fluorescence. We will present modifications to this potential which improve the quantitative fits to the observed Raman intensities. We also explore in more detail the origin of the broad fluorescence and the validity of various 'standard' assumptions such as the Condon approximation for this experimental case.