Geometric topology is not really a unique field in the way that algebraic topology and general topology are. Its various subareas may share something of a common feeling (and indeed an arxiv category), but are often too diverse to share any common basic techniques. These include, for instance:
Low-dimensional topology (classical knots, 3-manifolds, 4-manifolds, etc.)
Morse theory, simple homotopy theory and algebraic K-theory of spaces
Dimension theory (of separable metrizable spaces)
Topology of manifolds (surgery theory, codimension two knots, etc.)
Singularity theory (of smooth maps), and e.g. pictures of explicit sphere eversions
PL topology (block bundles, collapsing, bistellar moves, etc.)
Generalized manifolds, wild knots, etc.
Group actions on manifolds
Manifold structures (smoothing/trangulability and the Hauptvermutung; also Lipschitz structures, CD-manifolds, ... )
Embedding theory (smooth embeddings of projective spaces, PL embeddings of polyhedra, etc.)
I'm sure I forgot many important subjects here; even the structuring of this list is rather arbitrary. The point is, you will probably not get far with diving in some depth into geometric topology without being more specific. If unsure, try low-dimensional topology, which is anyway more than half of contemporary geometric topology by any count. Ryan and Jim gave some good suggestions in their math.SE answers.
There have been a few books mentioning 'geometric topology' in the title, but they are often specialized and/or advanced. For instance, the 'geometric topology' notes by Sullivan and Lurie are mostly (though not entirely) focused on manifold structures, and are firmly grounded in methods which are very clever and useful but kind of external to geometric topology (localization, Galois theory and simplicial sets).
Arguably, closer to the point are Fenn's 'Techniques of geometric topology' and Ferry's 'Geometric topology notes'. These two are certainly not equivalents of some canonical algebraic topology text such as Spanier's or Hatcher's but perhaps the closest to such an equivalent that I can think of.