A percentage of what?

Most definitions say relative humidity is "how much water is in the air compared to how much it could hold." That is correct and it is where the difficulty starts, because it slides past the only hard part: "could hold" is not a constant.

Air's capacity for water vapour is set almost entirely by its temperature, and it rises steeply — roughly doubling for every 10°C. So:

The most water air can hold, by temperature.
TemperatureCapacityWhat 50% would be
0°C / 32°F4.8 g/m³2.4 g/m³
10°C / 50°F9.4 g/m³4.7 g/m³
20°C / 68°F17.2 g/m³8.6 g/m³
30°C / 86°F30.3 g/m³15.1 g/m³

50% at 30°C is more than six times the water of 50% at 0°C. Same percentage, same "half full" — and the glass changed size.

Strictly, it is about pressure

The textbook definition is not in grams at all. Relative humidity is the ratio of the actual vapour pressure of the water in the air to the saturation vapour pressure at that temperature:

RH = 100 × e / es(T) Vapour pressure is the share of total air pressure contributed by water vapour.

The grams version above is a good mental model and very nearly the same number, but the pressure version is what instruments and formulas actually use. The formula guide works it through properly.

What the number is genuinely good for

Relative humidity has a bad reputation among people who have just learned about dew point, and it does not deserve one. It is the right measure whenever the question is about a rate or about a surface:

  • Evaporation speed. How fast washing dries, sweat evaporates or paint cures depends on how much room the air has left, which is exactly what the percentage measures.
  • Mould. The thresholds are defined in relative humidity at the surface, because what mould responds to is water activity, not grams per cubic metre. That is why the mould threshold is 60% and not a quantity.
  • Static. Below about 40% the microscopically thin water film on surfaces stops forming, charge stops leaking away, and you start getting shocks.
  • Comfort standards. ASHRAE 55 and the EPA guidance both specify ranges in percent.

What it is not good for is comparing two conditions. For that you need the quantity or the dew point.

Saturation is less water than you think

A useful piece of perspective: saturated air at 20°C is about 1.4% water by mass. Even in the tropics, water vapour is a small minority of what you are breathing.

The reason humidity matters so much despite being a trace component is that the change in that trace is what drives evaporation, and evaporation is how almost everything — bodies, laundry, paint, soil — sheds water.

Where the reading comes from

Modern instruments measure relative humidity directly with a capacitive sensor: a polymer film between two electrodes takes up water in proportion to the ambient relative humidity, changing its capacitance measurably. Dew point and absolute humidity are then computed from that reading and the temperature.

So the chain runs opposite to the intuition — the percentage is the measurement and the more "fundamental" figures are derived from it. How humidity is measured covers the other methods.