Amber suppression is a useful method of genetically incorporating a non-natural amino acid (NAA) into a protein during translation by utilizing an NAA-charged amber suppressor tRNA (sup-tRNA). A wheat germ extract (WGE) is suitable for this method by virtue of its high productivity and versatility in addition to its advantages as a cell-free translation system. However, in spite of this high potential, a genetic NAA incorporation system in WGE has not been sufficiently optimized in terms of sup-tRNAs, in contrast to that in E. coli and its cell extracts. We herein rationally optimized amber sup-tRNAs to efficiently incorporate a model NAA, p-acetyl-phenylalanine (AcPhe), into a protein in WGE, via flexizyme-based aminoacylation. The optimized sup-tRNA (named tLys-opt) that was pre-charged with AcPhe exclusively yielded up to 220 μg mL(-1) of AcPhe-incorporated protein (yellow fluorescent protein, YPet) under the optimal conditions. This high productivity is comparable to the best reported yield of a similar NAA-incorporated protein synthesized with an engineered aminoacyl-tRNA synthetase/sup-tRNA pair in WGE, despite the fact that tLys-opt that has released AcPhe was not reused at all in this study. The results clearly show both the necessity of optimizing sup-tRNAs for efficient NAA incorporation and the validity of our strategy for their optimization. Because the optimization strategy described here is expected to be applicable not only to amber sup-tRNAs for other NAAs but also to ones used in other acylation methods, it would facilitate the synthesis of large amounts of various types of NAA-incorporated proteins in WGE.