Merging is the action of joining one or more streams of MIDI data into one. Although conceptually simple this is a non-trivial task. A good analogy is a set of points on a railway line changing to allow trains from different lines onto one. The points have to be changed quickly to prevent trains waiting or getting another crash into it and if left in the wrong state a derailment occurs.

MIDI mergers are required wherever two sets of MIDI sources are required simultaneously at one input. Some specialized equipment may incorporate merging of MIDI data with its own data, e.g. timecode to clock synchronizers.

There are three common problems associated with merging:

The Output should never be routed back to an Input, if this happens data will circulate continuously, like a feedback howl, and often the only way of stopping it is to pull the cables out.

Merging two of the same sources will produce conflicting data or strange jumping e.g. two pitchbend wheels in different positions will cause the receiving equipment to jump to the latest position. Merging two MIDI clocks will produce a composite higher tempo clock and two sources of MIDI Time Code will produce unreadable time code. It is desirable to be able to selectively filter the individual inputs to prevent this type of conflict.

Merging too much data onto one MIDI cable may result in lost data as the data buffers in receiving equipment get overloaded or lagging delay effects. The latter is more noticeable with continuous controllers and these may need to be "thinned" or scanned at a lower rate at source.

Mergers such as the M-Audio Merge 2x2 are often placed in a position within a MIDI system where all data passes through and so have to work reliably under all conditions. If more than two sources are required to be merged it is better to use a merger that can handle that number of inputs. Cascading two input mergers will produce accumulative processing delays, which may be undesirable.