Radicool21's answer is only correct in the second-generation games. If there is both an accuracy modifier and an evasion modifier in the later games, the evasion modifier will be subtracted from the accuracy modifier, the result is capped at 6/-6 if it exceeds one of them, and the multiplier corresponding to that modified stat stage is used. In other words, the result is exactly the same whether the accuracy is -6 and the evasion is 6, the accuracy is -6 and the evasion is 0, or the accuracy is -3 and the evasion is 3.
So if the opponent's move's accuracy is 100, the modified accuracy will be simply 100 * 1/3 = 33. 1/3 is the multiplier corresponding to an accuracy/evasion stat stage of -6/6. You can find the rest in the Bulbapedia article Radicool21 linked to.
If you want it laid out in equations, first figure out the modified accuracy stat stage a:
a = min(6, max(-6, n - m))
Then the final accuracy of the move can be calculated as follows:
p * max(3, 3 + a)/max(3, 3 - a)