In order to reduce background signals in Q beta replicase-mediated bioassays, a target-dependent probe amplification strategy has been proposed that utilizes recombinant RNA hybridization probes that contain an inserted molecular switch. A molecular switch is an internal region of the probe that undergoes a conformational change when the probe hybridizes to its target. We investigated whether non-hybridized probes (which cause background signals) could be selectively destroyed by incubating the probe-target hybrids with ribonuclease III, which should cleave the non-hybridized probes and leave the hybridized probes intact. Two problems with this assay design were observed. First, ribonuclease III cleaved probe-target hybrids non-specifically when the target was an RNA, thereby destroying all of the bound probes. And second, the expected conformational change in the molecular switch did not occur when the probes were bound to their targets, apparently because the hairpin stem formed by the molecular switch was too long. Although these results demonstrated that the original assay design could not work, they provided insights that have led to better designs for target-dependent amplification assays. In these assays, the probes will be DNA molecules containing short-stemmed molecular switches. Non-hybridized probes will be selectively destroyed by incubation with a restriction endonuclease.