Blocking is a well-known technique of experimental control. Pair matching is simply a special case of blocking. Pair matching is a form of local control, which controls for extraneous variables by pairing up similar subjects before applying analysis.
Before employing pair matching, an experimenter first defines the extraneous variables. After doing so, she can match pairs of subjects on their extraneous variables. The experimenter, in essence, reorganizes the data into pairs of subjects that are similar (constant on the extraneous variables).
After matching the pairs, the experimenter can then randomize treatments. This is done by randomizing one subject in a pair to a certain treatment, and giving the other subject in the pair the other treatment. Employing this method on all pairs allows the experimenter to analyze the data with local control.