In this talk we will discuss a randomized Kaczmarz method for linear inverse problems. We will explain why randomized Kaczmarz is suitable for linear inverse problem by providing pre-asymptotic analysis and studying the structure of some specific problems. Numerical examples validated our analysis.