During the 1970s and 80s, Schmid and others analyzed in great detail how a polarized variation of Hodge structure degenerates near a normal crossing divisor. All subsequent applications of Hodge theory in algebraic geometry rely to some extent on their work. In the talk, I am going to outline a simplified proof for most of Schmid's results; I hope this will clarify the meaning of the "limiting mixed Hodge structure", the Hodge norm estimates, and other things.