Abstract
A model for RNA secondary structure is studied that has a rugged energy landscape with many alternative local minima. Several features are observed that are qualitatively similar to the replica theory of spin glasses. Numerical results suggest that the overlap distribution remains broad and non-self-averaging for long random RNA sequences at low temperatures. Significant ultrametric correlations are observed in the distances between triplets of states. Clustering algorithms are used to illustrate the hierarchical arrangement of low energy states in configuration space.