Lustre (1984) and AltaRica (2000) are two languages used to describe critical systems. The first one is a data-flow programming language; and its main purpose is the writing of formally verified programs implemented on embedded hardware. The second language has been designed mainly for the modelling and the analysis of non-specific systems. These languages are supported by different toolboxes and software workbenches which motivates the development of translators between the two languages. In this report we describe a way to translate Lustre models into AltaRica ones (and vice versa). With some restrictions on the input model, the translation produces a model whose semantics is very closed to the original. This work has been granted by the CERT-ONERA in the context of the ISAAC European Project.