At that time, Florida was the only state that paid a private company to purge the voter file of ineligible voters, in effect allowing a private company to make the administrative decision of who is not eligible to vote.
The State of Florida's Division of Elections was required to contract with a private entity to purge its voter file by chapter 98.0975 of the Florida statutes, which had been enacted by the Florida legislature to address voter registration fraud found during the 1997 Miami mayoral election, according to the United States Civil Rights Commission Report on 2000 Florida Elections.
Previously voter purging had been conducted (sometimes controversially) by local elections officials. During the US Civil Rights movement of the 1960s, local elections officials in southern states, including Florida, were the subjects of lawsuits, marches and civil disobedience as African-Americans attempted to register to vote. This led to the passage of the federal 1965 Voting Rights Act banning discriminatory practices that kept African-Americans off the voter rolls.
The first firm hired on 1998 to purge the voter rolls was Professional Service Inc., which charged $5,700 for the job. Later in the same year, the state placed an open request for tenders to bid for the job. The contract was assigned to DBT Onlines, despite the fact that its bid was the highest-priced. The state gave the job to DBT for a first year fee of US $2,317,800 with total fees eventually reaching US $4 million The Florida Department of Elections terminated Professional Service Inc.'s contract in 1999. DBT Online was later acquired by ChoicePoint, of Atlanta, in early 2000.
In February 2000, in a phone conversation with the BBC's London studios, ChoicePoint vice-president James Lee said that the state "wanted there to be more names than were actually verified as being a convicted felon".

Lee went on saying that the state then ordered DBT to shift to an even lower threshold of 80% match, allowing also names to be reversed (thus a person named Thomas Clarence could be taken to be the same as Clarence Thomas). Besides this, middle initials were skipped, Jr. and Sr. suffixes dropped, and some nicknames and aliases were added to puff up the list.
"DBT told state officials", testified Lee, "that the rules for creating the [purge] list would mean a significant number of people who were not deceased, not registered in more than one county, or not a felon, would be included on the list. DBT made suggestions to reduce the numbers of eligible voters included on the list". According to Lee, to this suggestion the state told the company, "Forget about it".
"The people who worked on this (for DBT) are very adamant... they told them what would happen", said Lee. "The state expected the county supervisors to be the failsafe." Lee said his company will never again get involved in cleansing voting rolls. "We are not confident any of the methods used today can guarantee legal voters will not be wrongfully denied the right to vote", Lee told a group of Atlanta-area black lawmakers in March 2001.
The first list DBT Online provided to the Division of Elections in April 2000 contained the names of 181,157 persons. Approximately 65,776 of those included on the first list were identified as felons.
In May 2000, DBT discovered that approximately 8,000 names were erroneously placed on the exclusion list, mostly those of former Texas prisoners who were included on a DBT list that turned out never to have been convicted of more than a misdemeanor. Later in the month, DBT provided a revised list to the Division of Elections (DOE) containing a total of 173,127 persons. Of those included on the "corrected list", 57,746 were identified as felons.
Examples:

Voter demographics authority David Bositis, a senior research associate at the Joint Center for Political and Economic Studies in Washington, DC, reviewed The Nation's findings and concluded that the purge-and-block program was "a patently obvious technique to discriminate against black voters". He noted that based on nationwide conviction rates, African-Americans would account for 46% of the ex-felon group wrongly disfranchised.
Between May 1999 and Election Day 2000, two Florida secretaries of state, Sandra Mortham and Katherine Harris, distributed the scrub lists produced by the cleansing process to counties and ordered the 57,700 people identified as "ex-felons" to be removed from voter rolls. Together the lists comprised nearly 1% of Florida's electorate and nearly 3% of its African-American voters.
At the time of the election, the purge list contained a number of false positives — people identified as felons who were not actually felons.

Nearly 3,000 out-of-state ex-felons with voting rights restored, as well as voters linked to felonies in states which do not remove felons from voting rolls or that automatically restore voting rights, were included on the list. According to a 1998 ruling by the 2nd District Court of Appeals, they cannot then be ruled ineligible by another state.
DBT had decided in March of 1999 not to include felon lists from South Carolina or Texas, which automatically restore voting rights, but that was overruled by the head of the Florida Office of Executive Clemency, Janet Keels, who ordered inclusion of any felon who did not have a written order of clemency, even from these states, wrongly placing 996 voters on the felon list. Florida did not restore their voting rights until three months after the election.
Additionally, a number of persons listed as felons had been convicted of misdemeanors only, and therefore were eligible by law.
Greg Palast, who has investigated this issue and identified occurrences of these problems, provides a sample of 23 names as they appear on the Florida 2000 felons list, with five examples of these erroneous listings highlighted (this represents a minimum rate of inaccuracy of 22% in this sample). Thomas Cooper, the second one in the list, was listed as being convicted on January 30, 2007.
The Georgia Bureau of Investigation has said that the national figure for "standard" margin of error for legal disenfranchisement is about 2%.
Database experts consulted by Greg Palast (including DBT's vice-president) told him that in order to obtain 85% accuracy or better, one needs at least the following three things:
ChoicePoint, in contrast, used virtually no Social Security numbers, did not check address histories, and used no database cross-checking, although it had 1,200 databases that could be employed for the task.
Because some of the source databases used did not list race, the matching criteria did not require a match with the voter's race for inclusion in the felon list. However, the decision was also made to enlarge upon this decision, and rule as ineligible the voter in question even if there was an explicit disagreement between the races listed on the source database and the voter list. According to the Palm Beach Post, more than 1,300 registered voters were matched with felons although their races or sexes were different. 
Mark Hull, the former senior programmer for CDB Infotek, a ChoicePoint company, said the state and ChoicePoint could have chosen criteria that would have brought down the number of false positives to less than 1%. George W. Bush officially received fewer than 600 more votes in Florida than Al Gore (2000 election).
The only reliable measure of accuracy of the felon list comes from Leon County (Tallahassee), whose in-house experts checked each name in their county one by one. Out of the 694 named felons in Tallahassee, they could verify only 34 of them, or 5%.
The Palm Beach Post reported that

Additionally, there are other accuracy problems with the list. For example, Linda Howell, Madison County supervisor of elections, who is not a convicted felon and was never on the felon list provided by the Division of Elections or the Florida Department of Law Enforcement, erroneously received a form letter referencing a prior felony conviction from the Florida Department of Law Enforcement stating:
Ms. Howell recalled, "I had sent the letter to one of my voters and he sent in the verification form. Instead of picking up his name, they picked up my name and sent me the information."
Some voters on the list did not receive advance notice that they were ineligible to vote until they appeared at the polls. Some had even received new voters cards in the mail.
Phyllis Hampton, general counsel of the Florida Elections Commission, testified that her office could investigate the wrongful removal of a Floridian from the voter rolls if there was evidence of a willful violation. Ms. Hampton stated:
People like Washington County Elections Chief Carol Griffen have argued that Florida was in violation of the National Voter Registration Act of 1993 "by requiring those convicted of felonies in other states (and subsequently restored their rights by said states), to request clemency and a restoration of their rights".
On February 17, 2001, the American Civil Liberties Union of Florida, in collaboration with Florida Legal Services and the Florida Justice Institute, launched the Equal Voting Rights Project. 
In June 2001, The U.S. Commission on Civil Rights released a report and statements
arguing that Florida was, on numerous counts, in violation of the Voting Rights Act of 1965 and recommending:
On February, 2002, the NAACP and four other groups filed suit against Harris (NAACP v. Harris), a former state election chief, and the county elections supervisor.
The lawsuit cites the state, several counties and the contractor over procedures for voter registration, voter lists and balloting. The suit charges that Black voters were disenfranchised during the 2000 presidential election, and argued that Florida was in violation of the Voting Rights Act of 1965 and the US Constitution's Equal Protection Amendment. The parties reached a settlement wherein ChoicePoint will donate $75,000 to the NAACP and will reprocess the voter file on the plaintiffs' terms.
In August, 2002, the U.S. Commission on Civil Rights released a briefing summary regarding the progress of voting rights in Florida