The original effect
The original Hanany–Witten transition was discovered in type IIB superstring theory in flat, 10-dimensional Minkowski space. They considered a configuration of NS5-branes, D5-branes and D3-branes which today is called a Hanany–Witten brane cartoon. They demonstrated that a subsector of the corresponding open string theory is described by a 3-dimensional Yang–Mills gauge theory. However they found that the string theory space of solutions, called the moduli space, only agreed with the known Yang-Mills moduli space if whenever an NS5-brane and a D5-brane cross, a D3-brane stretched between them is created or destroyed.
They also presented various other arguments in support of their effect, such as a derivation from the worldvolume Wess–Zumino terms. This proof uses the fact that the flux from each brane renders the action of the other brane ill-defined if one does not include the D3-brane.
Furthermore they discovered the S-rule, which states that in a supersymmetric configuration the number of D3-branes stretched between a D5-brane and an NS5-brane may only be equal to 0 or 1. Then the Hanany-Witten effect implies that after the D5-brane and the NS5-brane cross, if there was a single D3-brane stretched between them it will be destroyed, and if there was not one then one will be created.
More generally, NS5-branes and D5-branes may form bound states known as (p,q) 5-branes. The above argument was extended in spontaneous supersymmetry breaking.
Dual forms of the effect
Via a series of T-dualities one obtains the result that in any type II superstring theory, when an NS5-brane and a Dp-brane cross one necessarily creates of destroys a D(p-2)-brane. Lifting this statement to M-theory one finds that when two M5-branes cross, one creates or destroys an M2-brane. Using S-duality one may obtain transitions without NS5-brane. For example, when a D5-brane and a D3-cross one creates or destroys a fundamental string.