"The electoral math is less solid for Clinton than it was for [President Barack] Obama four years ago," Silver said in an interview with George Stephanopoulos on ABC News' "This Week."
Silver became a prominent election forecaster during the 2008 election, and his assessment of candidates' electoral chances have been closely followed ever since. This year his work has come under scrutiny because while his models show Clinton the favorite in the race, he is less bullish on her chances than other forecasters.
The reason Clinton's polling lead is less safe than Obama's was four years ago is that the race is closer in some swing states, especially in the Midwest. That means that if she loses only one big state forecast for her, she loses the White House, Silver said.
The other issue is that Clinton's support doesn't come as close to 50 percent as Obama's did four years ago. When she's down around 44 percent, Silver said, that leaves her vulnerable to a surge of undecided voters breaking for Trump and tipping the scales for him.
"If Clinton were to beat her polls by 3 points ... you see something we call a borderline landslide. But if it goes the other way ... all of a sudden Trump could very easily win the Electoral College," he said.
A 3 point lead is not safe enough, in Silver's model.
The FiveThirtyEight editor-in-chief defended his projection model from critics who say it should take into account early-voting data.
"I would be a little bit careful," he warned of those numbers. "Democrats also in 2014 told themselves a lot of stories about how they would be saved by the early vote and got wiped out across the board in the midterm [congressional elections]."