This article presents a technique for nonlinear devices characterization and testing by identification of the Volterra kernels. After presenting the concept of Volterra modeling, we present a CAT tool based on the Volterra modeling technique. Then we apply the technique for the case-studies of a nonlinear system and a purely nonlinear MEMS cantilever. By finding the Volterra kernels of each of the case-studies we show how they can be used for characterization and test. Through this work, we introduce and discuss the multi-level MLS stimulus generation and response processing necessary to calculate the Volterra kernels. The multi-level stimulus can be generated easily on-chip and the response processing can also be simplified to be implemented on-chip for the purpose of BIST. The use of pseudorandom testing for MEMS characterization has never been addressed in the past to the best of our knowledge.