Roy G. Saltman, the federal government’s leading expert on computerized voting whose overlooked warning about the vulnerability of punch-card ballots presaged the hanging chad fiasco in Florida that came to symbolize the disputed recount in the 2000 presidential election, died on April 21 in Rockville, Md. He was 90.

His death, in a nursing home, was caused by complications of recent strokes, his grandson Max Saltman said.

In a 132-page federal report published in 1988 and distributed to thousands of local voting officials across the country, Mr. Saltman, an analyst working for the National Institute of Standards and Technology, cautioned that the bits of cardboard that voters were supposed to punch out from their ballots, known as chads, might remain partly attached (hence, hanging), or pressed back into the card when the votes were counted.

Either event would render the voter’s choice uncertain or, if the ballot appeared to be picking more than one candidate, invalid.

“It is recommended,” Mr. Saltman said flatly, “that the use of pre-scored punch card ballots be ended.”

His recommendation was largely ignored, certainly in Florida, where the initial count in the 2020 election gave the Republican candidate, Gov. George W. Bush of Texas, a 1,784-vote lead over the Democrat, Vice President Al Gore, a margin so close that state law required a recount.

Armies of lawyers and political operatives descended on Florida, suits and countersuits were filed, and recounts were started and stopped in various counties. The spectacle of election workers examining punch-card ballots through magnifying glasses, to try to determine a voter’s intent, popularized the term hanging chad as it raised doubts about the accuracy of the count.

After five weeks of recounts, the U.S. Supreme Court stepped in on Dec. 12, 2000, and, in a 5-to-4 decision, stopped a state court-ordered recount, with Mr. Bush holding a 537-vote lead over Mr. Gore. Florida’s 25 Electoral College votes, and the presidency, were awarded to Mr. Bush.

“It has always puzzled me why my report never got a wider acceptance,” Mr. Saltman told USA Today in 2001. “It takes a crisis to move people, and it shouldn’t have.”

The counting crisis that crippled the presidential transition in 2000 prompted congressional hearings that led in 2002 to the Help America Vote Act, which outlawed the use of punch cards in federal elections.

As recently as last month, Fox News agreed to pay $787.5 million to resolve a defamation suit filed by Dominion Voting Systems after Fox TV personalities falsely claimed that Dominion’s voting machines were susceptible to hacking and had switched votes in the 2020 election from President Donald J. Trump to Joseph R. Biden Jr. The company’s patents cite Mr. Saltman’s early reports on punch-card vulnerabilities as proof that Dominion’s voting technology had overcome those flaws.

As early as 1976, Mr. Saltman warned that “we have a serious problem of public confidence in computers and a serious problem of public confidence in public officials, and around election time they tend to coalesce.”

When his bosses at the federal agency discounted his early concerns, Mr. Saltman got a $150,000 grant to study voting mishaps around the country.

He found a report that reviewed Detroit’s first punch-card voting experience in a 1970 primary election. It turned up “design inadequacies of the voting device” that had invalidated ballots because voters had unintentionally voted for more than the prescribed number of candidates. Similar concerns about punch-card voting were raised after a 1984 election for property appraiser in Palm Beach County, Fla.

In 1988, Mr. Saltman’s prescient report, “Accuracy, Integrity and Security in Computerized Vote Tallying,” recommended banning the pre-scored punch-card voting machines that would create the counting crisis in Florida in 2000.

He also recommended against the use of computer systems that would prevent voters from examining their ballots for accuracy before leaving the polls, and that would not produce an immediate printed paper trail for election officials to examine in a recount.

“The defects in the pre-scored punch card voting system are fundamental and cannot be fixed by engineering or management alterations,” Mr. Saltman wrote. He added that “manual examination of pre-scored punch card ballots to determine the voter’s intent is highly subjective.”

“For example,” he continued, “manual counters are forced to determine whether a pinprick point on a chad demonstrated an intent to register a vote.”

Max Saltman said his grandfather had expressed concern that nearly all electronic voting systems in the United States still relied on complex operating systems, despite his warnings about their vulnerabilities.

Charles Stewart III, an M.I.T. professor of political science who consulted with Mr. Saltman, said by email: “Roy appreciated how computers could help to make election administration better, by automating vote counting, which is a very tedious and error-prone exercise when done by hand. But, he demonstrated that these machines sometimes broke down, and it was foolish not to design systems that took this fact into account.”

Roy Gilbert Saltman was born on July 15, 1932, in Manhattan to Ralph Henry Saltman, a son of immigrants from Russia, and Josephine (Stern) Saltman, who had immigrated from Budapest as an infant. His father was a production manager in the garment industry and later at an electrical appliance factory. His mother was a homemaker.

Raised in the Bronx and in Sunnyside, Queens, Roy graduated from Brooklyn Technical High School.

He earned a degree in electrical engineering from Rensselaer Polytechnic Institute in Troy, N.Y., in 1953. In 1955, he received a master’s in engineering from M.I.T., where he worked on the guidance systems for the Nautilus, the first nuclear submarine. He also studied engineering at Columbia University and was granted a master’s degree in public administration from the American University in Washington in 1976.

In 1969, after jobs at Sperry Gyroscope Co. and IBM, he joined the Department of Commerce’s National Institute of Standards and Technology, where he worked on software policy and served on the U.S. Board on Geographic Names, the agency charged with maintaining the uniform usage of geographic names within the federal government.

His first marriage, to Lenore Sack, ended in divorce. In 1992, he married Joan Ettinger Ephross. She died in 2008.

In addition to his grandson Max, he is survived by his sons, David and Steven, and a daughter, Eve, from his marriage to Dr. Sack; his stepchildren, David, Peter and Sara; two other grandchildren; and six step-grandchildren.

After he retired in 1996, Mr. Saltman became an election consultant.

The belated attention his reports received after the 2000 election, in part as a result of his testimony to the House Committee on Science in May 2001, prompted him to write what became a definitive book, “The History and Politics of Voting Technology” (2006).

He also continued to speak out on election issues. In a letter to The Washington Post in 2005, he warned that Georgia’s requirement that voters have a photo ID card, at a cost of $20 every five years, might violate the Constitution’s prohibition of a poll tax.

As Sue Halpern wrote in The New Yorker in 2020, plenty of potential problems with electronic voting machines that Mr. Saltman identified remain: “tallies that can’t be audited because the voting machines do not provide a paper trail, software and hardware glitches, security vulnerabilities, poor connections between voting machines and central tabulating computers, conflicts of interest among vendors of computerized systems, and election officials who lack computer expertise.”

Mr. Saltman often said that there was no margin of error in voting, that civic engagement and confidence in the electoral system was too vital to a democracy to leave any grounds for misgivings.

“An election is like the launch of a space rocket,” he often said. “It must work the first time.”