Since both the stereochemical rules and structural knowledge can be expressed as geometric constraints, the macromolecular structure modeling and prediction problems can be seen as constraint satisfaction problems (CSP) in 3-D space. The backtracking algorithm in the first versions of MC-SYM (Macromolecular Conformations by SYMbolic programming) searches the conformational space of RNA such that all generated models are consistent with a given set of input constraints (sound), and all possible constructions are returned by the procedure (complete). The RNA description and geometric constraints are entered in ASCII scripts. The nucleotide and atomic coordinates are computed dynamically using the information provided in the script. The conformational space explored is determined by the choice of pre-computed nucleotide conformations and transformations. The "default" database provided with MC-SYM can easily be extended or completely redefined. New nucleotide conformations are introduced in the conformational database by defining PDB-like formatted coordinates. New transformations are implemented by introducing the matrices and lists of atoms from which the transformations must be applied. The time required by MC-SYM and the number of models generated depend on the amount of structural information used by the search procedure to prune the conformational space. The models produced by MC-SYM are of atomic precision (contain all atoms) and can be computed on UNIX workstation-type computers.