A model for the secondary structure of the self-cleaving RNA from hepatitis delta virus was tested. Specific base changes were introduced in each of four regions with the potential for base-pairing (stems I-IV), and for each variant sequence, a rate constant for cleavage was determined. In each stem, mutations that would interfere with Watson-Crick base-pairing also reduced the first-order rate constants by 10-10(4)-fold relative to the unmodified version. Within stems I and II and a shortened form of stem IV, compensatory changes resulted in rates of cleavage equal to or greater than the unaltered ribozyme sequence. Stem III compensatory mutants cleaved faster than the uncompensated mutants although they were not as active as the natural sequence, suggesting additional sequence-dependent requirements within this region. Structure probing of RNA containing the stem II mutations provided an independent confirmation of stem II in the ribozyme. The predictive value of the model was tested by designing two trans-acting ribozymes which were circularly permuted composites of genomic, antigenomic, and unique sequences. The core of these two catalytic RNAs was the same, but they otherwise differed in that, in one of them, a constraining tetraloop sequence was added to stem II. Both ribozymes catalyzed the trans cleavage of a substrate oligoribonucleotide, thus providing additional evidence for stem II and the proposed structure in general.