Gas path fault diagnosis plays an important role in guaranteeing safe, reliable and cost-effective operation for gas turbine engines. Measurements selection is among the most critical issues for diagnostic method implementation. In this paper, an integration approach for optimal measurements selection, which combines finger print diagrams analysis, health parameters correlation analysis, performance estimation uncertainty index analysis and fault cases validation based on genetic algorithm, has been proposed and applied to assess the health condition of a two-spool split flow turbofan in test bed. First, mathematical description of an engine gas path fault diagnosis process was given and the influence coefficient matrix was also calculated based on a well calibrated nonlinear engine performance simulation model. Second, the number of combination candidates was reduced from 782 to 256 and three measurements were picked out using the finger print diagrams analysis and the health parameters correlation analysis. Then, the number of the combination candidates was further narrowed down to 13 using the performance estimation uncertainty index analysis. A nonlinear genetic algorithm fault diagnosis method was applied to test the diagnostic ability of the remaining measurement candidates. Finally, an optimal measurement combination was worked out which demonstrated the effectiveness of the integration approach. This integration approach for optimal measurements selection is also applicable to other type of gas turbine engines.