Looking at both successful (and unsuccessful) attempts to crack ancient undeciphered writing systems over the course of history, the aim of this paper is to study how decipherment can contribute to a general theory of writing. It is argued that two conditions have to be fulfilled for decipherment to be possible; it has to be acknowledged: (1) that writing is a system possessing an autonomous internal organization, and (2) that the relationship between this system and the linguistic system cannot be envisioned as an equivalence, as a translation or as a transcoding, but has to be seen as a synchronization or adjustment. This synchronization is based either on parallelism or on symbolization and distribution. The decipherment of Egyptian hieroglyphs is a good illustration of the first option, as it relies on a comparative approach that resorts to analogy. In the second case (closer to cryptography), internal reconstruction is favored and the formal regularities of the system are the starting point. Ventris’ decipherment of Linear B is a famous example of such an approach. As such, decipherment can be envisioned as an interpretative process that allows one to (re)construct a system, from its substance to its form, and the systematic character of the decipherment is the guarantee of its validity.