RMS stands for root mean square. It expresses the average energy contained in a given stretch of a waveform, letting you look at the sustained strength of a sound separately from peak, which shows the instantaneous maximum.
Why not a simple average?
An audio waveform swings positive and negative around zero, so averaging it directly makes the values cancel out and approach zero even for a loud sound. RMS squares the values, averages them, and takes the square root, so it expresses the energy in the waveform regardless of sign.
Difference from peak
A peak meter catches brief large excursions, which makes it good for preventing clipping. An RMS meter averages over a period, so it responds less to short peaks and shows more clearly how strong a sound is on a sustained basis. Watching both side by side gives you a sense of the difference between attack and density.
Readings differ between meters
The time window used for the RMS calculation, the update rate, and the reference all vary from meter to meter. The same audio will not necessarily read the same number. How the meter is calibrated with respect to the peak and RMS values of a sine wave also produces differences, so use the same settings when comparing.
Difference from LUFS
RMS is a mathematical average of the waveform’s energy. LUFS measures perceived loudness using weighting that reflects the ear’s frequency sensitivity, measurement windows suited to the purpose, and gating. RMS is still a clue to how loud something feels, but LUFS is better suited to matching the perceived loudness of different material.
Using it in production
A sound with a higher RMS at a similar peak level appears to have a strong, dense sustained portion. When comparing before and after a compressor or comparing several pieces of material, do not judge from RMS alone; use peak, LUFS, dynamic range, and your ears together. The loudness adjustment tool lets you check peak and LUFS.
