Structurally diverse ensembles of intrinsically disordered proteins or regions are difficult to determine, because experimental observables usually report a conformational average. Therefore, in order to infer the underlying distribution, a set of experiments that measure different aspects of the system is necessary. In principle, there exists a set of cross-correlated relaxation (CCR) rates that report on protein backbone geometry in a complementary way. However, CCR rates are hard to interpret, because geometric information is encoded in an ambiguous way and they present themselves as a convolute of both structure and dynamics. Despite these challenges, CCR rates analyzed within a suitable statistical framework are able to identify conformations in structured proteins. In the context of disordered proteins, we find that this approach has to be adjusted to account for local dynamics via including an additional CCR rate. The results of this study show that CCR rates can be used to characterize structure propensities also in disordered proteins. Instead of using an experimental reference structure, we employed computational spectroscopy to calculate CCR rates from molecular dynamics (MD) simulations and subsequently compared the results to conformations as observed directly in the MD trajectory.