Well tests performed in fractured rocks often display non-classical drawdown behaviors due to strong hydraulic heterogeneity and various flow structures investigated. Such signals are thus challenging to interpret but contain important information on flow-bearing structures (i. e., hydraulic heterogeneity and connectivity), particularly useful for site understanding and modeling. We propose here to interpret a large-scale pumping test performed in a major fault zone developed in an inter-bedded sandstone shale sequence of turbidite origin. The pumping test was carried out over 31 days and responses were recorded up to 725 m distance from the pumping well. Successive and different flow regimes are observed at the pumping well and are consistent with channel type flow structures. As a guideline, synthetic well tests were simulated for several idealized channel configurations. This numerical investigation shows that similar drawdown responses may be obtained for different channel configurations and properties, which highlight the problem of uniqueness and uncertainty on hydraulic properties. Such problems may be partly reduced with the number of observation wells properly distributed and geological information to constrain the interpretation of a single realization. The interpretation of the real experiment is then proposed consistent with drawdown observations and geological structures, and limitations are discussed based on available information.