Controversy
The EJ25 head-gasket question, honestly unresolved.
Ask any Subaru forum and you'll hear it stated flatly, like a law of physics: the 2.5-liter boxers eat head gaskets. The pattern turns up everywhere you look — everywhere except the one kind of document that would make it official. This is a case file with a strong lead and no smoking gun, and we're not going to pretend otherwise.
Two failure patterns, both reported
The story actually comes in two chapters. The 1996–1999 first-generation "EJ25D" engines are reported to fail internally — a head gasket letting go between the cooling passages, coolant vanishing, the car quietly cooking itself. The later EJ251/EJ252/EJ253 engines are reported to fail the other way, weeping externally around the 100,000-mile mark. Different mechanisms, same reputation, both told and retold across the naturally-aspirated Legacy, Outback, Forester, Impreza and Baja.
Where the trail goes cold
Follow the claim back and it thins out. What's underneath it is repair-trade-press retrospectives — shops writing up what they see on lifts — not a Subaru technical service bulletin, not a class-action filing, not an independent engineering teardown. That's real-world signal, and worth respecting. It is also a weaker tier of evidence than this site will use to stamp a specific failure rate as fact. So the reputation stays exactly that: a reputation.
What would crack the case
Three things would move this from lead to verdict, and none has been pulled yet: a primary Subaru TSB naming the affected engines, a direct query of NHTSA's complaint database, or Consumer Reports' own reliability-survey data. Until one of them is on the table, the honest answer is the interesting one — everybody knows, and nobody here can prove it.