Sanctions fail most of the time at their headline goals: in the largest historical study, 204 sanctions episodes produced at least partial success in 34 percent of cases, and only 31 percent of the 80 episodes aimed at regime change. But "failed" is the wrong frame — the 2021–2026 record shows sanctions consistently doing real work at raising a target's costs, even when they never change its government.
What does the evidence actually say?
The most-cited answer comes from the Peterson Institute's Economic Sanctions Reconsidered (3rd edition), which scored 204 sanctions episodes from 1914 to 2006. At least partial success: 34 percent overall. Split by goal, the pattern is blunt — modest and limited objectives such as releasing a political prisoner succeeded 51 percent of the time, while attempts at regime change and democratization succeeded in 31 percent, disrupting military adventures in 21 percent, and other major policy changes in about 30 percent (PIIE preview chapter). Success correlated with modest goals, strong trade links, high cost to the target, and cooperation among the senders — not with ambition (PIIE preview chapter 4).
The number deserves a second reading. Robert Pape's re-analysis of the study's earlier 40 claimed successes (covering 1914–1990) found only 5 clear cases where sanctions themselves caused the outcome; 18 were determined by military force, 8 were outright failures, 6 were trade disputes, and 3 were indeterminate (Pape, International Security, 1997). His point survives thirty years of argument: when a sanctions episode "works," ask what else was in the room — almost always, force or a coup. Since 1970, US unilateral sanctions in particular have fallen below 20 percent success (Hufbauer et al., via ETH case survey).
What did Russia's sanctions change — and not change?
Russia is the cleanest test case of the century so far, and the results cut both ways. The regime is untouched and the war did not stop — but the sanctions did not miss either. The EU embargo and the $60 G7 price cap, effective from December 2022, initially cut the discount on Russian Urals crude to a peak of about $38–40 per barrel below Brent in January 2023; Russia's own shadow-fleet spending then clawed the discount back down to roughly $13–18 by 2024 (KSE Institute, July 2024). Dallas Fed researchers concluded in May 2024 that the price cap alone did little to lower Russian export prices — roughly half the price discount came from the embargo forcing longer, costlier reroutes to India and China, and the rest from those buyers' market power (Dallas Fed, May 2024).
The evasion is quantified, not anecdotal: by June 2024, 89 percent of Russia's seaborne crude moved on shadow-fleet tankers outside Western services, and cumulative lost oil revenue from the sanctions was estimated at $78.5 billion for December 2022 through June 2024 — real money, at a cost of a war that continued regardless (KSE Institute, July 2024). A 2025 peer-reviewed assessment reached a similar judgment: the import embargo was the consequential sanction, the price cap functioned less as a price tool than as a partial transport embargo, and total Russian losses were conservatively at least 0.9 percent of GDP (Energy Policy, 2025).
What did the Iran record actually settle?
Iran is the textbook case of sanctions as a regime-shaping instrument rather than a regime-changing one. Decades of sanctions did not change Iran's government, and the 2015 nuclear deal that lifted them — the JCPOA — held only as long as every party stayed in it. After the US withdrew in 2018 and Iran breached its limits, the European E3 triggered the JCPOA's snapback mechanism on 28 August 2025, and the pre-2015 UN Security Council resolutions came back into force at midnight GMT on 28 September 2025 — ten years after they had been lifted (E3 joint statement, September 2025; EU Council, September 2025).
Snapback is the rare sanctions instrument that worked mechanically — unanimity was not required, and the sanctions simply returned. What it did not do was produce the concessions the E3 sought before the deadline: Iran refused IAEA access to its sites and an accounting of its high-enriched uranium stockpile, and talks at the UN General Assembly in September 2025 produced nothing (Reuters, September 2025). Ten years on, the mechanism restored the legal framework in a day and changed Iran's behavior not at all.
If they rarely work, what do they actually do?
| Case | What sanctions were asked to do | What they did instead (as of September 2026) | |---|---|---| | Russia, 2022– | Reverse or deter the invasion of Ukraine | War continues; oil losses of ~$78.5bn over 18 months and a permanent shadow-fleet overhead do constrain war finance at the margin (KSE Institute) | | Iran, 2015–2025 | Verifiably cap the nuclear program | The deal's machinery snapped back cleanly in September 2025, but Iran had already breached the limits and the inspections were suspended (E3 statement) | | Türkiye, 2018 | Free a detained American pastor | Sanctions on two ministers and doubled steel tariffs in August 2018 were one accelerant among several in a lira collapse of roughly 40 percent against the dollar in 2018; Brunson was released in October 2018 — a modest-goal success that worked because the demand was narrow (CRS report, August 2018; BBC, August 2018) |
The Türkiye row is the honest exception that proves the rule. The sanctions did coincide with the pastor's release within weeks — but the currency crisis that made Ankara bleed was mostly homegrown, and the Congressional Research Service's assessment at the time placed US sanctions as an accelerant of an existing crisis, not its cause (CRS, August 2018). Narrow demand, willing target, cheap compliance: that is the entire recipe for a sanctions success, and it is rare.
Why do sanctions keep being used, then?
Because the alternatives are worse and the accounting is not only about compliance. The modest-goal record — 51 percent success — is genuinely respectable, and it covers most sanctions ever imposed: asset freezes on specific officials, arms embargoes, visa bans, extradition demands. The failure zone is the ambitious end — changing a government's mind about territory, survival, or its own regime (PIIE).
There are also the jobs sanctions are never billed for but reliably perform. They signal resolve to allies at a price cheaper than war. They degrade a target's military procurement — Russia's arms exports halved between 2014–18 and 2019–23, from 21 percent to 11 percent of the world market, before sanctions were even the main story (SIPRI, March 2024). They force the target to spend real resources on evasion — Russia's shadow fleet is a standing overhead, a tax it pays every day. And they make the point to third parties that the coercer can and will act — which is its own deterrent, regardless of the target's concession.
What would change this assessment?
Three things, none of them in evidence as of September 2026. A genuinely enforceable price cap — one where evasion is punished faster than the shadow fleet can grow — would turn the price cap from a political symbol into an actual price instrument (Dallas Fed, May 2024). Multilateral sanctions with real secondary enforcement would remove the leak that has let every major target since 2014 find a buyer. And senders could aim at the goals the data supports — narrow, verifiable, reversible demands — instead of regime change, where two generations of evidence put the success rate under a third and Pape's recount puts it near 5 percent of cases (Pape, 1997). Until then, the honest summary stands: sanctions are a tax, not a lever — they rarely compel, they often cost, and their only reliable products are the ones nobody put on the banner.