Like many people (I'd guess most people reading this blog), I had a Rubik's Cube in the 80s.
The only way I could solve it was looking up moves in the book You Can Do The Cube by Patrick Bossert. (Patrick was only 12 when he wrote the book—he's now an entrepreneur and management consultant; see his home page).
In high school I had Rubik's Magic which I took detailed notes on and came up with a solution on my own. Later I came across a book with a much shorter solution that I seem to recall memorizing in the bookstore without buying the book.
Also in high school, I had Rubik's Clock which was also easy enough to solve on my own.
Then last year, when I was in Walt Disney World surprising my sister for her 21st, I bought a Rubik's Cube again with the goal of learning how to solve it.
Wikibooks has a nice page on How to solve the Rubik's Cube that has a straightforward algorithm as well as a discussion of various other approaches favoured by speedcubers.
Wikipedia also has a very interesting article on Optimal solutions for Rubik's Cube. It was only August this year that Kunkle and Cooperman proved Twenty-Six Moves Sufﬁce for Rubik’s Cube (pdf of paper).
UPDATE (2008-03-27): Now see Twenty-Five Moves Suffice
The original post had 3 comments I'm in the process of migrating over.