Previous studies have described poliovirus genomes in which the internal ribosome entry (IRES) for encephalomyocarditis virus (EMCV) is positioned between the P1 and P2-P3 open reading frames of the poliovirus genome. Although these dicistronic poliovirus genomes were replication competent, most exhibited evidence of genetic instability, and the EMCV IRES was deleted upon serial passage. One possible reason for instability of the genome is that the dicistronic genome was at least 108% larger than the wild-type poliovirus genome, which could reduce the efficiency of encapsidation. To address this possibility, we have constructed dicistronic poliovirus replicons by substituting the EMCV IRES and the gene encoding luciferase in place of the poliovirus P1 region; the resulting dicistronic replicons are smaller than the wild-type poliovirus genome. One dicistronic genome was constructed in which the poliovirus 5' nontranslated region was fused to the gene encoding luciferase, followed by the complete EMCV IRES fused to the P2-P3 region of the poliovirus genome (PV-Luc-EMCV). A second dicistronic genome, EMCV-Luc-PV, was constructed with the first 108 nucleotides of the poliovirus genome fused to the EMCV IRES, followed by the gene encoding luciferase and the poliovirus IRES fused to the remaining P2-P3 region of the poliovirus genome. Both dicistronic replicons expressed abundant luciferase following transfection of in vitro-transcribed RNA into HeLa cells at 30, 33, or 37°C. The luciferase activity detected from PV-Luc-EMCV increased rapidly during the first 4 h following transfection and then plateaued, peaking after approximately 24 h. In contrast, the luciferase activity detected from EMCV-Luc-PV increased for approximately 12 h following transfection; by 24 h posttransfection, the overall levels of luciferase activity were similar to that of PV-Luc-EMCV. To analyze encapsidation of the dicistronic replicons, we used a system in which the capsid protein (P1) is provided in trans from a recombinant vaccinia virus (VV-P1). The PV-Luc-EMCV replicon was unstable upon serial passage in the presence of VV-P1, with deletions of the EMCV IRES region detected even during the initial transfection at 37°C. Following serial passage in the presence of VV-P1 at 33 or 30°C, we detected deleted genomes in which the luciferase gene was fused with the P2-P3 genes of the poliovirus genome so as to maintain the translational reading frame. In contrast, the EMCV-Luc-PV replicon was genetically stable during passage with VV-P1 at 33 or 30°C. The encapsidation of EMCV-Luc-PV was compared to that of monocistronic replicons encoding luciferase with either a poliovirus or EMCV IRES. Analysis of the encapsidated replicons after four serial passages with VV-P1 revealed that the dicistronic replicon was encapsidated more efficiently than the monocistronic replicon with the EMCV IRES but less efficiently than the monicistronic replicon with the poliovirus IRES. The results of this study suggest a genetic predisposition for picornavirus genomes to contain a single IRES region and are discussed with respect to a role of the IRES in encapsidation.