Many complex notions and constructions admit non-Archimedean analogs, so it is very natural to seek for a non-Archimedean analytic geometry parallel to the classical complex one. A naive attempt to reformulate complex definitions does not lead to a reasonable theory because of bad properties of non-Archimedean topology: for example, C_p is totally disconnected so it cannot be the underlying topological space of an analytic affine C_p-line. . In the sixties, Tate developed rigid analytic geometry, where this difficulty is overcome by imposing a Grothendieck topology on (what is defined to be) analytic spaces. Thus, set-theoretically analytic spaces are what one would expect them to be, but the topology is highly non-trivial and sometimes counter-intuitive, especially in not quasi-compact cases. . Another approach to non-Archimedean geometry was discovered by Berkovich in late eighties. One saturates Tate's rigid spaces with additional points (not surprisingly, these are certain points of the rigid site). The new spaces possess a nice topology: they are locally compact (if Hausdorff), locally linearly conected and (as a new and difficult result of Berkovich states) are locally contractible. It makes them very useful for some applications, including theories of fundamental groups and etale cohomologies. . In the series of three lectures, I'm going to give an introduction to the theory of Berkovich analytic spaces. The first lecture will be devoted to the class of good spaces (a notion which is natural in analytic geometry but is not so easy to define in rigid geometry). This class contains separated curves and analytifications of schemes. In the second lecture we will study analytic curves and their local and global structure, in particular the analytic semi-stable reduction theorem. Finally, the third lecture will be devoted to defining general analytic spaces and relations between analytic geometry and other theories: rigid, formal and adic geometries.