Nowadays, most of turbine engine tests are processed using an open loop, i.e. the measurements are verified and treated a posteriori, sometimes weeks or months after the end of the test. The scope of the present project is to develop a new methodology which enables real time detection of faulty measurements and the suppression of the source of these faults during the test. This implies the development of a method which is sufficiently robust to cope with a lot of faulty data (up to 20 to 30 % of the measurements). Most of the existing methods make use of a least square approach so that a set of parameters is sought which minimizes the RMS error between the measurements and the values provided by a numerical model of the installation. This type of approach is efficient in finding whether a set of measurements is valid but cannot locate the fault(s). In this paper an alternative is proposed based on a distribution of the measurement noise introduced by Huber. This so-called robust validation method has been tested on a single spool, single flow and variable geometry nozzle turbojet. Both the least square and Huber's function approaches have been tested and compared in terms of efficiency. The robust estimator based on Huber's error function has shown to be much more powerful for both finding an invalid set of measurements and locating the faulty measurements.