In recent years empirical data that convey information on RNA structure have become available at increasing rates. This includes in particular methods for chemical probing and well chemical crosslinking. Although these sources of information are not sufficient to determine RNA structures accurately by themselves, they can drastically improve the quality of model-based structure predictions. In this presentation I will describe the including of chemical probing data into RNA folding algorithms in the form of pseudo-energies and compare this to closely related approaches using phylogenetic variation instead. Crosslinking data only provide approximate locations of base pairs. Nevertheless, it turns out that this information can drastically improve the quality of predicted structures.