The 2000 US presidential election between Al Gore and George W. Bush has been the most intriguing and controversial in American history. Using the Florida ballot data, Wu showed that the 2000 election result could have been reversed had the "butterfly ballot effect" been eliminated. Through a combinatorial approach, Harger concluded that Gore should have won the election based on certain assumptions. In this paper, the author analyses the voting results of the 2000 election by applying the multiple regression technique to a set of national voters data and discusses the prediction error under three different voting systems. The study indicates that party affiliation was the dominant factor in the 2000 election and that a proportional voting system was more predictable than popular and electoral systems. The 2000 US presidential election source data is appended. (Contains 5 tables.)