Key features of an ideal RNA-based vaccine against coxsackievirus B3 (CVB3) are (i) limited genome replication/virus production (to minimize vaccine-related pathology) and (ii) abundant virus protein synthesis (to maximize immunogenicity). These attributes may apply to CVB3 RNAs lacking up to 250 nucleotides (nt) from their 5' terminus; these RNAs do not give rise to infectious progeny, but they have been reported to retain the entire CVB3 IRES (mapped to nt approximately 432-639) and to produce large quantities of viral protein in transfected cells. Here, we constructed five 5' RNA deletion variants that, to our surprise, failed to protect against CVB3 challenge. We investigated the reasons for this failure and conclude that (i) a 5' terminal deletion as short as 32 nt abolishes CVB3 RNA replication in transfected cells; (ii) this deleted RNA, and others with longer deletions, do not direct abundant protein synthesis in transfected cells, probably as a consequence of their replicative incapacity; and (iii) the CVB3 IRES is substantially larger than previously thought, and its 5' boundary lies between residues 76 and 125, very closely approximating that of the poliovirus IRES.