ICE in 2026: How Popular It Really Is, Whether Its Tactics Are Legal, and How the Numbers Compare With Past Presidents

Two Polls, Same Country, Opposite Answers

In January 2025, AP-NORC asked Americans about deporting immigrants convicted of violent crimes. About 8 in 10 said yes — 82% of Democrats, 86% of Republicans ^[4]. A Harvard-Harris poll taken in late May 2026 found nearly the same thing: 80% support, including 71% of Democrats and 90% of Republicans ^[14]. That looks like as close to consensus as American politics gets.

Then a Siena College poll of New York State voters, taken the same January as the AP-NORC survey, asked a narrower question. Deport someone with no criminal record, who has lived here a long time? Support fell to 39%, with 42% opposed ^[16]. Same country, same general topic, opposite headline. Both results are real. Neither one is "the" public opinion on deportation — they are answers to two different questions wearing the same name.

The pattern holds beyond who gets named. It also holds for what gets named. Pew found that in June 2025, 54% of Americans disapproved of increasing ICE workplace raids ^[18]. A separate Pew survey in January 2026 found 61% think ICE officers should not be allowed to hide their faces with masks ^[59]. ABC News and Ipsos found Americans opposing ICE's tactics by 62% to 31% ^[19]. So approval of the goal — remove people who committed crimes — and approval of the methods being used to pursue it are two different numbers, and they move in opposite directions.

Opinion has also swung over time, and in both directions. Gallup found 55% of Americans wanted immigration reduced in 2024, the highest in two decades ^[12]. By June 2025, that had fallen to 30%, and a record 79% called immigration a good thing for the country ^[13]. Support among Hispanic adults for the administration's immigration approach fell especially hard, to 21% in Gallup's June 2025 poll, versus 35% nationally ^[13]. Reuters/Ipsos found a similar drop ^[32], and Fox News found 59% of voters now call ICE "too aggressive," up 10 points since July 2025 ^[33]. So whichever side of this argument someone is on, they can find a real poll that backs them up. That is not spin. It is the actual shape of the data.

The Officials Are Not Their Voters

Part of why this argument never resolves is that the people arguing loudest — elected officials, party activists, donors — often sit to one side of their own voters. This runs in both directions. Democratic officials in many places have taken positions on sanctuary policy well to the left of the median Democratic voter. Republican officials have pushed positions on mass deportation and legal-immigration cuts that run ahead of, or in some cases behind, where their own voters actually sit.

The machinery producing this gap is structural, not personal, and it works the same way for both parties. Closed primaries with low turnout let the most motivated, most ideological voters pick nominees. Donors and advocacy groups shape which positions get funded and which get punished. Safe seats remove the incentive to court the middle. Media and fundraising both reward sharp, quotable positions over ambivalent ones. None of this is a claim that either party is unusually captured — it is a description of an incentive structure that pulls both parties away from their own base, in whichever direction their activist wing leans.

Who Signs the Warrant Decides the Fight

Move from opinion to law, and the first thing to do is separate three different questions that keep getting mashed into one. What do current statutes and Supreme Court rulings actually permit? What have courts specifically ruled on real 2025 and 2026 practices? And what is still open, moving through litigation, with no answer yet? Treating those as one question is where most of the confusion in this debate comes from.

Start with what is settled. The government's basic power to detain and remove people who are in the country illegally rests on real statutes and real Supreme Court cases. Congress gave immigration officers the power to make warrantless arrests of people believed to be here illegally and likely to flee ^[1]. The Supreme Court upheld mandatory detention of noncitizens convicted of certain crimes in Demore v. Kim in 2003 ^[2], and in Jennings v. Rodriguez in 2018 read the detention statutes as not requiring periodic bond hearings ^[3]. Fixed immigration checkpoints don't need individual suspicion of any one driver ^[9]. And immigration law generally gets more deference from courts than ordinary domestic law enforcement — a principle called the plenary power doctrine, reaffirmed in Trump v. Hawaii in 2018 ^[10]. The idea is that decisions about who may enter or stay in the country are treated as close to the core of national sovereignty, so courts give the political branches more room than they would in, say, a police-search case. That doctrine is why immigration law so often looks like it plays by different constitutional rules than everything else — because, to a real degree, it does.

But "the government may remove someone" is a different sentence from "an agent may force open a door to do it." That is where the fight actually lives, and it usually comes down to one distinction: who signed the paper. An ICE administrative warrant — the forms known as I-200 and I-205 — is signed by the agency itself, not a judge. It authorizes arrest, not entry into a home. A judicial warrant, by contrast, carries a neutral judge's independent finding of probable cause. The Supreme Court held in Payton v. New York in 1980 that only a judicial warrant, not an administrative one, can authorize forcing entry into someone's home ^[38]. That single distinction is doing most of the work behind the disputes over home entry, workplace raids, and detainers.

Three camps argue this out, and each one has a real case. The Justice Department, ICE leadership, the officers' union, and restrictionist groups like the Center for Immigration Studies argue the legal foundation is simply settled: the statutes, Demore, Jennings, and the checkpoint cases are real law, not aspiration, and roughly 80% of the public backs deporting people convicted of crimes — a number that held up even through 2025's broader backlash ^[4][14]. This camp can also point to a real, on-the-merits court win: in June 2026, a federal appeals court upheld the expanded fast-track removal process, a tool used by administrations of both parties since 1996 ^[15][11]. Its blind spot, by its critics' account, is that the popularity finding covers only the narrow thing it names — actual tactics like raids and masked agents poll badly even among Republicans — and a popular goal doesn't answer whether a specific method is lawful ^[18][19].

Immigrant-rights litigators — the ACLU, the National Immigration Law Center, RAICES, and allied groups — argue that the "who signed the warrant" distinction is exactly the problem, and that 2025 enforcement shifted hard toward people with no criminal record at all. Government data compiled by TRAC found 74.2% of ICE detainees in January 2026 had no criminal conviction ^[23]. This camp also points to district-court rulings that found roving immigration stops likely unconstitutional ^[20]. Its critics note that several of its biggest 2025 wins were later stayed or reversed by higher courts, meaning the disputed tactics remain legally permitted for now, and that the June 2026 appeals ruling was a genuine government win on the merits, not just a technical delay ^[15][21].

A third camp — libertarian and some conservative Fourth Amendment scholars at Cato, the Institute for Justice, and Reason's Volokh Conspiracy — accepts the government's basic power to remove people but argues the mechanics themselves cross constitutional lines, including for U.S. citizens caught up in the process. Even a real judicial warrant for one person can't authorize entering someone else's home without a separate search warrant, under a 1981 case called Steagald v. United States ^[50] — which matters directly when a citizen shares a house with someone ICE is after. In January 2026, a federal judge in Minnesota ruled that agents violated the Fourth Amendment by battering down a door using only an administrative warrant, exactly the scenario this camp had been warning about ^[53][54]. In December 2025, the Supreme Court also blocked the government from using federalized National Guard troops for immigration enforcement in Illinois, finding no legal authority for it ^[55]. This camp's own weak point, by its critics' account, is that its biggest warning — that these tools will spread beyond immigration into ordinary policing — is a prediction, not a ruling. Courts have treated immigration as its own separate constitutional territory for more than a century ^[10].

The Deportation Math Nobody Agrees On

Now the history, which is easy to state and easy to misread. Barack Obama's administration carried out about 3.1 million formal removals over eight years, and by 2014 immigrant-rights groups were calling him the "deporter-in-chief" ^[6][35]. Donald Trump's first term removed fewer people in total than Obama's did ^[7]. On the raw numbers, the current administration's second-term pace — around 150,000 in the first six months, on track to top 300,000 for the year — would be the highest annual figure since 2014 ^[8].

But the composition changed underneath those totals, and that composition is really what the argument is about. Obama's interior arrests — the kind of neighborhood and workplace raids people picture when they hear "deportation" — peaked at 188,422 in fiscal year 2011, then fell more than 63%, to about 69,473, by 2015 ^[36]. Much of his later, still-large totals came from the border, not from the interior. In 2025, by contrast, TRAC found 74.2% of ICE detainees had no criminal conviction, and in one late-2025 window, only about 3% of new arrests involved anyone with a conviction ^[23]. So two administrations can post comparable headline totals while enforcing very differently: one increasingly at the border, the other increasingly inside American communities, against people with no criminal record.

One more claim gets repeated often enough to test directly: that earlier presidents deported people at scale with no real public outcry. That claim doesn't hold up. The #Not1More campaign, organized through the National Day Laborer Organizing Network, launched in April 2013 ^[34]. In March 2014, the president of the National Council of La Raza — now UnidosUS — publicly called Obama "deporter-in-chief" ^[35]. The outcry existed. Whether it was smaller, differently organized, or just less visible in a pre-2020s media environment than today's protests is a separate question the evidence doesn't settle.

What Nobody Knows Yet

Several of the biggest legal questions here simply have no answer yet. The Supreme Court paused a lower-court block on ethnicity-informed street stops in September 2025 — but a pause is not a ruling, and the Fourth Amendment question underneath it remains open ^[21]. Only one district court, in Minnesota, has ruled on the merits of the administration's May 2025 memo authorizing forced home entry on agency paperwork; no appeals court has weighed in yet, and the ruling can still be appealed ^[53]. Whether 2025 was genuinely the deadliest year for ICE detainees on a per-person basis, not just in raw deaths, is unclear too — the detained population itself grew by roughly half over the same period, and final federal inspector-general figures haven't been published ^[29].

Perhaps the plainest gap: the government keeps no public, audited count of how many U.S. citizens it has detained. ProPublica documented more than 170 such cases over about nine months, including nearly 20 children — but that number is a journalistic floor, not an official statistic, because no one is officially counting ^[47][48].

Across the spectrum, each outlet tends to reach for the number that fits its lens rather than manufacturing new facts. Restrictionist commentary leans on the broadest, most favorable crosstabs and the phrase "committed crimes" rather than "convicted" ^[14]. Libertarian legal writers frame the fight as a Fourth Amendment problem untethered from immigration politics either way ^[39][41]. Immigrant-rights advocacy leads with the tactics and the poll swing toward opposition, saving the durable criminal-conviction consensus for later in the piece, if it appears at all ^[16]. Nonpartisan pollsters and watchdog groups, by contrast, tend to split the question apart rather than reduce it to one number — goal versus tactic, interior versus border ^[18][30]. That may be the most honest response available: not a single verdict, but a map of exactly where the real disagreement sits.