Abstract—One of the most significant drawbacks that the previously proposed watermarking schemes are suffered from, is size of the watermark picture which should be of limited length. One possible way out of this bottleneck is going into molecular layers. In this regard, We present the first watermarking model in DNA computing. This scheme can be implemented on an in-vivo computer. The proposed scheme uses the bacteriophage lambda genome for coding watermark image (owner-related information) and the Escherichia coli (E. coli) bacteria genome for coding the host image. The presented coding scheme is based on the silent mutation of codons which produces harmless bacteria and phages. The significant advantage of this scheme over other methods is that the process of embedding the watermark image in the host image takes place naturally in the process of infecting E. coli cells using bacteriophage lambda in lysogenic cycle. The owner of the watermark image can easily extract his information using the centrifuge technique from the resulting test tube of the proposed scheme. A computer simulation is also conducted in this paper that depicts the processes of coding, embedding and extraction to demonstrate the biological procedures involved in this paper.