A double-precision floating-point value is stored using two components: the significand (also called mantissa or significant digits) which contains the actual digits, and the exponent which determines the scale or magnitude. This representation allows efficient storage of very large or very small numbers.