The paper introduces ARWEN, a software program designed to detect tRNA genes in metazoan mitochondrial nucleotide sequences. Mitochondrial tRNA genes often exhibit degenerate sequences and structures compared to their bacterial ancestors, making current tRNA gene prediction programs ineffective. ARWEN employs a heuristic algorithm to identify tRNA genes with a high detection rate, achieving nearly 100% sensitivity for annotated genes. The program is user-friendly, with limited parameter settings and easy-to-interpretable results. It is available online and can be downloaded for testing. ARWEN outperforms tRNAscan-SE in detecting true tRNAs, with a significantly lower rate of false positives. The combination of ARWEN and tRNAscan-SE is recommended for improving the annotation of tRNAs in metazoan mitochondrial sequences.The paper introduces ARWEN, a software program designed to detect tRNA genes in metazoan mitochondrial nucleotide sequences. Mitochondrial tRNA genes often exhibit degenerate sequences and structures compared to their bacterial ancestors, making current tRNA gene prediction programs ineffective. ARWEN employs a heuristic algorithm to identify tRNA genes with a high detection rate, achieving nearly 100% sensitivity for annotated genes. The program is user-friendly, with limited parameter settings and easy-to-interpretable results. It is available online and can be downloaded for testing. ARWEN outperforms tRNAscan-SE in detecting true tRNAs, with a significantly lower rate of false positives. The combination of ARWEN and tRNAscan-SE is recommended for improving the annotation of tRNAs in metazoan mitochondrial sequences.