Abstract. In inverse problems it is quite usual to encounter equations that are ill-posed and require regularization aimed at finding stable approximate solutions when the given data are noisy. In this paper, we discuss definitions and concepts for the degree of ill-posedness for linear operator equations in a Hilbert space setting. It is important to distinguish between a global version of such degree taking into account the smoothing properties of the forward operator, only, and a local version combining that with the corresponding solution smoothness. We include the rarely discussed case of non-compact forward operators and explain why the usual notion of degree of ill-posedness cannot be used in this case.Key words. Degree of ill-posedness, regularization, linear operator equation, Hilbert space, modulus of continuity, spectral distribution, source condition.AMS subject classifications. 47A52, 65J20.1. Introduction. It is an intrinsic property of a wide class of inverse problems that small perturbations in the data may lead to arbitrarily large errors in the solution. Hence, abstract models of inverse problems are frequently associated with operator equations formulated in infinite dimensional spaces that are ill-posed in the sense of Hadamard. For their stable approximate solution such equations require regularization when the given data are noisy. The mathematical theory and practice of regularization (see, e.g., the textbooks [1,4,7,11,20,25] and the papers [2,5,9,22,24,26,28,33,35]) takes advantage of some knowledge concerning the nature of ill-posedness of the underlying problem. This nature regards available a priori information and the degree of ill-posedness from which conclusions with respect to appropriate regularization methods and efficient regularization parameter choices can be drawn.We restrict our considerations here on ill-posed linear operator equations