Masking means that one puts in some "noise" in the opposite ear while testing an ear. The reason to do this is to prevent sound from the side being tested from going over to the good side This is called the "cross-over problem". See the illustration below for an example of the cross-over problem.
When to do masking --
If one is doing pure-tone testing, the masking signal should be narrow band noise centered on the test frequency. If one is doing speech, one should use speech noise.
Masking is not always needed. When doing air conduction audiograms, this cross-over problem occurs to a different extent depending on the earphone being used. Earphones that go around the ear run into cross-over problems when the sound is about 40 db or louder. Earphones that go into the ear (inserts) have problems at about 60 db. Thus if the ears are both normal, masking is not needed. One should always mask air conduction if there is a 40db or greater threshold.
For bone conduction testing, the sound always goes to both ears, and one cannot test one ear by itself (usually) without using masking. The best strategy therefore, for bone conduction testing, is to always mask.
Example of situation where masking is necessary for correct results.
When masking is not used is that one may incorrectly conclude that there is better hearing in a "bad" ear, than is really the case, because without masking, some of the sound is actually appreciated by the good ear.
This person is completely deaf on the right side. However, because masking was not used, the bone lines (top) make it appear that they could have normal hearing.
This is the same person as above, tested this time with masking. Here it is clear that the patient cannot hear on the right side. This is the correct result.
This situation occurs when there is a conductive type hearing loss in both ears, which is moderate to severe. The dilemma is that an adequate intensity to mask the non-test ear crosses over to the testing ear and invalidates the thresholds.
Enough masking is too much masking. There are several "work arounds" to this. When one is masking air conduction, if one can (i.e. no perforation) use insert earphones.
Other methods include using bone-ABR testing, ABR is unaffected by the central effect of masking but the sound still stimulates both ears and also one is limited by one's bone stimulator. Another method is the SAL (sensorineural acuity level) technique. In the SAL method, both ears are masked equally through an oscillator placed over the forehead. This is done by performing air conduction thresholds while using bone masking, and then doing some adjustments for the "air conduction shift". This method involves a large number of subtractions of noisy values, and for this reason seems (to us) to be an intrinsically unreliable (and unwieldy) method.
The bottom line is that when one encounters the "masking dilemma", usually this is simply noted on the audiogram and one accepts that it is impossible to determine the details of the hearing loss.