Well-established linear procedures for detection and identification of control system failures are extended to nonlinear systems within a fuzzy logic framework. This rule-based approach has particular advantages for application to nonlinear systems that can be represented as linear parameter varying (LPV) systems. The effectiveness of the technique in the presence of modeling uncertainties, actuator dynamic failures and failures in redundant actuators is evaluated. It is pointed out that for certain practical failures, identification of the exact failure is a limitation that existing approaches to failure detection and identification do not address. Use of additional sources of information and a knowledge-based diagnostics is highlighted. Simulations conducted on a nonlinear aircraft model are included.