A very common question. It is to do with the way Excel stores
fractional values, as most decimal fractions cannot be represented
accurately in binary. Take a simple decimal example, using 2 dp:

x = 1/3 = 0.33

x + x + x = 0.33 + 0.33 + 0.33 = 0.99

which is not equal to 1.

Pete

On Feb 4, 5:08*pm, JasonG wrote:
If I put these values and formulae, I get a zero total, which is what I expect

a1: 123.456
a2: 234.567
a3: 358.023

=a3-a2-a1 * *(gives zero exactly)

If I put this, I get a rounding error:

=(a3-a2-a1) * (gives 1.42e-14)

In another spreadsheet, the error was much larger (2e-9).

Excel should give the same result either way, but it doesn't.

Does anyone know why?

Jason