The inverse problem of ocean color consists in deriving the inherent optical properties (IOP) of marine particles from a reflectance spectrum measured at the sea surface. Such a problem is ill-posed or ambiguous because of the nonuniqueness of the solution; that is, several combinations of IOP values can lead to a unique reflectance spectrum. Currently, great efforts are made in the development of inverse methods to accurately retrieve the IOPs. However, many fewer studies have been devoted to the analysis of the ambiguities, which affect yet the error on the IOPs retrieval. In this paper, the ambiguities related to the ocean color problem in coastal waters are characterized and their implications for inverse modeling are studied. A synthetic data set is created on the basis of radiative transfer modeling. The simulations are constrained using in situ observations and statistical rules to make the data set realistic. The ambiguity rate of remote sensing reflectance (Rrs) spectra is around 90%, thus meaning that the ocean color problem is extremely ambiguous. The influence of the ambiguities on the IOPs retrieval is evaluated. It is demonstrated that the error that is ascribed to the occurrence of ambiguity is equal to the dispersion of all the plausible IOPs solutions. The ambiguity error made on the total absorption coefficient is shown to be greater in highly absorbing water mass. On the other hand, the ambiguity error made on the total backscattering coefficient is higher in turbid scattering waters. Finally, different strategies to reduce the effects of ambiguities are discussed.