Every picornavirus studied thus far has a sequence within the 5'-non-coding region that is required for internal ribosome binding and translation of the polyprotein. In an attempt to identify this region in hepatitis A virus we constructed a truncated hepatitis A virus (HAV) cDNA clone that contains the entire 736 bp 5' non-coding region (5'-NCR) and 754 base pairs of the viral capsid coding region (P1) under control of the SP6 promoter. In vitro transcription and translation of this transcript in a rabbit reticulocyte lysate yielded a protein product of about 29 kDa as analyzed by autoradiography following sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). A series of mutations of this construct have defined a minimal sequence between bases 347 and 734 in the 5'-NCR that is required for efficient in vitro translation. The deleted constructs (D 523-734 and D 632-734) showed a reduced ability to translate in the rabbit reticulocyte lysate system in comparison with the full-length 5'-NCR construct, pH1489. The translation of these deleted constructs was artificially restored by the addition of a 5'-terminal methylated cap structure, m7GpppG, to the RNA. This increase in translational efficiency could be competed away with cap analog (m7GDP) thus indicating that this region is required for cap-independent internal ribosome binding for HAV translation.