The number of digits to which numbers are represented. For example, if p bits are allocated to the mantissa in the representation of floating-point numbers used in a particular computer, then in that computer floating-point numbers have p bits of precision. In general, the precision of floating-point numbers is proportional to their value (i.e. relative) whereas the precision of fixed-point numbers is absolute (independent of the value).
It is important not to confuse the term precision with accuracy. For example, the number
has eight-decimal digit precision, irrespective of what it represents. If this number represents 22/7 then it is also accurate to eight decimal digits but if it represents the irrational number π then it is accurate only to three decimal digits.