Structural RNA modules play central roles as architectural organizers and sites of ligand binding in RNA molecules, and are recurrently observed in RNA families throughout the phylogeny. Here we describe a computational approach for identifying known 3D structural modules in single and multiple RNA sequences. We applied a Bayesian Network model for sequence identification and a thermodynamic model for candidate filtering. If an alignment of homologs sequences is available the results of each individual sequence can be clustered together to improve the specificity of the algorithm. We applied this tool to the discovery of four well known modules: G-bulge loop, kink-turn, C-loop and tandem-GA loop. In control test sequences we found all of the known modules with a false discovery rate of 0.23. Scanning through 1,444 publicly available alignments, we identified 21 yet unreported modules and 141 known modules. This was the first example of a structural module detection on RNA sequences and RMDetect can be used to refine RNA 2D structure, assemble RNA 3D models, and search and annotate structured RNAs in genomic data.