Repeat-induced gene silencing (RIGS) establishes the centromere structure, prevents the spread of transposons and silences transgenes, thereby limiting recombinant protein production. We previously isolated a sequence (B-3-31) that alleviates RIGS from the human genome. Here, we developed an assay system for evaluating the influence of repeat sequences on gene expression, based on in vitro ligation followed by our original gene amplification technology in animal cells. Using this assay, we found that the repeat of B-3-31, three core sequences of replication initiation regions (G5, C12, and D8) and two matrix attachment regions (AR1 and 32-3), activated the co-amplified plasmid-encoded d2EGFP gene in both human and hamster cell lines. This upregulation effect persisted for up to 82 days, which was confirmed to be repeat-induced, and was thus designated as a repeatinduced gene activation (RIGA). In clear contrast, the repeat of three bacterial sequences (lambda-phage, Amp, and ColE1) and three human retroposon sequences (Alu, 5'-untranslated region, and ORF1 of a long interspersed nuclear element) suppressed gene expression, thus reflecting RIGS. RIGS was CpG-independent. We suggest that RIGA might be associated with replication initiation. The discovery of RIGS and RIGA has implications for the repeat in mammalian genome, as well as practical value in recombinant production.