If ribozymes are to be exploited in vivo, it is necessary to select ribozymes that are functional in the intracellular environment. Ribozymes selected in the intracellular environment should retain their function in vivo as well as in vitro. We have devised a novel system for selection of active ribozymes from pools of active and inactive ribozymes using the gene for dihydrofolate reductase (DHFR) as a selective marker. In our first attempt, a sequence encoding either an active or an inactive ribozyme was connected upstream of the gene for DHFR. Each plasmid was designed such that, when the ribozyme was active, the ribozyme would cleave the target site and, as a result, the rate of production of DHFR would be high enough to endow resistance to trimethoprim (TMP). However, a critical defect may be associated with introduction of a ribozyme upstream of the DHFR gene because, during actual screening for active ribozymes on the 5' side from a pool of random sequences, there is the danger of selecting sequences that are not related to the activity of ribozymes. Indeed, some upstream linker sequences affected the level of expression of the DHFR protein and, as a result, the resistance of Escherichia coli to TMP. Therefore, we newly constructed a 3'-connected ribozyme system, and activities in vivo of 5'-connected and 3'-connected ribozymes were compared. We found that the cleavage efficiencies in vivo were nearly identical for the two types of ribozyme, 24% for the 5'-side ribozyme and 23% for the 3'-side ribozyme, indicating that polysomes did not seem to inhibit the action of the 3'-connected ribozyme. In both cases, when cells were transformed with a 1 : 1 mixture of active and inactive ribozyme-coding plasmids, it was mainly the cells that harbored the active ribozyme that survived in the presence of TMP.