Gore. In recounts after the inaugration, if there had been a statewide recount, Gore would have won.
Where do you get this information from? The recounts done by the NY Times (hardly a bastion of Republicanism) showed Bush the winner under the various counting methodologies.
Media post-electoral studies/recounts
After the election, USA Today, The Miami Herald, and Knight Ridder commissioned accounting firm BDO Seidman to count undervotes, that is, ballots which did not register any vote when counted by machine. BDO Seidman's results, reported in USA Today [7], show that under the strictest standard, where only a cleanly punched ballot with a fully removed chad was counted, Gore won by three votes. Under all other standards, Bush won, with Bush's margin increasing as looser standards were used. The standards considered by BDO Seidman were:
Lenient standard. Any alteration in a chad, ranging from a dimple to a full punch, counts as a vote. By this standard, Bush won by 1,665 votes.
Palm Beach standard. A dimple is counted as a vote if other races on the same ballot show dimples as well. By this standard, Bush won by 884 votes.
Two-corner standard. A chad with two or more corners removed is counted as a vote. This is the most common standard in use. By this standard, Bush won by 363 votes.
Strict standard. Only a fully removed chad counts as a vote. By this standard, Gore won by 3 votes.
The study remarks that because of the possibility of mistakes, it is difficult to conclude that Gore was surely the winner under the strict standard. It also remarks that there are variations between examiners, and that election officials often did not provide the same number of undervotes as were counted on Election Day. Furthermore, the study did not consider overvotes, ballots which registered more than one vote when counted by machine.
The study also found that undervotes break down into two distinct types, those coming from punch-card using counties, and those coming from optical-scan using counties. Undervotes from punch-card using counties give new votes to candidates in roughly the same proportion as the county's official vote. Furthermore, the number of undervotes correlates with how well the punch-card machines are maintained, and not with factors such as race or socioeconomic status. Undervotes from optical-scan using counties, however, correlate with Democratic votes more than Republican votes. Optical-scan counties were the only places in the study where Gore gained more votes than Bush, 1,036 to 775.
A larger consortium of news organizations, including the USA Today, the Miami Herald, Knight Ridder, the Tampa Tribune, and five other newspapers next conducted a full recount of all ballots, including both undervotes and overvotes. According to their results, under stricter standards for vote counting, Bush won, and under looser standards, Gore won. [8] However, a Gore win was impossible without a recount of overvotes, which he did not request.
According to the study, only 3% of the 111,261 overvotes had markings that could be interpreted as a legal vote. According to Anthony Salvado, a political scientist at the University of California, Irvine, who acted as a consultant on the media recount, most of the errors were caused by ballot design, ballot wording, and efforts by voters to choose both a president and a vice-president. For example, 21,188 of the Florida overvotes, or nearly one-fifth of the total, originated from Duval County, where the presidential race was split across two pages. Voters were instructed to "vote every page". Half of the overvotes in Duval County had one presidential candidate marked on each page, making their vote illegal under Florida law. Salvado says that this error alone cost Gore the election.
Including overvotes in the above totals for undervotes gives different margins of victory:
Lenient standard. Gore by 332 votes.
Palm Beach standard. Gore by 242 votes.
Two-corner standard. Bush by 407 votes.
Strict standard. Bush by 152 votes.
In 2003, United States citizens living in the state of Florida were asked who they voted for in the 2000 Election as part of the Statistical Abstract Census. The results showed President Bush receiving more than 1000 votes more than former Vice President Gore. However this result was badly tarnished when it was discovered that the man responsible for this census had links to the original Bush campaign in 2000.
This is excerpted from
http://en.wikipedia.org/wiki/2000_Presidential_Election