A reading of the ending of Taxi Driver
When the ending doesn’t work, instinct tells you the problem is there. You rewrite the last scenes. You adjust the climax. You give it more weight, more clarity, more impact. You change the order of events. You add a scene to close what seems open. And for a moment it seems to improve.
It doesn’t.
What you’re doing is intervening on the surface of something that was already broken long before. You’re repainting a wall with structural damage. The wall can look better. But the structure keeps failing.
Billy Wilder put it with a precision that still unsettles: “If you have a problem with the third act, the real problem is in the first act.” It’s not a paradox or a veteran screenwriter’s quip. It’s a structural diagnosis about how a story actually works. The ending is not the place where you decide how it ends. It’s the place the story arrives at when it can no longer keep avoiding what it has been building. If that place doesn’t exist, if the story hasn’t built anything that needs to arrive somewhere, then the ending is arbitrary. And an arbitrary ending is always felt, even if the audience can’t name it.
The difference between an ending that feels manufactured and one that feels inevitable is not in the closing scene. It’s in everything before it. In whether each scene has accumulated something with no way out. In whether the pressure has grown without artificial relief. In whether the story has been honest with its own internal logic from the beginning, or has been making deals with the expectations of the reader, the producer, the market, the fear.
Inevitability is not written. It accumulates. And when it hasn’t accumulated, there’s no way to write it at the end.
Paul Schrader wrote Taxi Driver from a place that was neither academic nor technical. It was personal. The isolation he describes in Travis Bickle, that inability to connect with the world around him, that sense of being on the margins of something everyone else seems to understand effortlessly, Schrader knew it from the inside. And that matters because it’s what makes the film work not as a thriller, not as social commentary, not as a character study in the sense the manuals describe. It works as a drift. And a drift has no resolution because it has no destination. It has direction. And that direction, from the first image, is inward. Toward the very core of the character.
Travis Bickle does not move toward any evident external objective. There’s no tangible goal he must reach, no antagonist he must defeat, no relationship he must repair. What there is instead is something else. An accumulation. Each scene doesn’t bring him closer to something, it sinks him a little deeper into himself. The city doesn’t change. The people around him don’t change. Betsy doesn’t change. Senator Palantine doesn’t change. The only thing that changes is the way Travis perceives all of it, and that perception becomes progressively more distorted, more rigid, more dangerous. What begins as discomfort becomes repulsion. What is repulsion becomes judgment. What is judgment becomes mission.
Scorsese shoots that drift without concessions. There’s no real scene of relief. No moment where the pressure genuinely eases. Every time the story seems to open toward something, toward Betsy, toward Iris, toward the possibility of connection, Travis’s awkwardness closes it immediately. Not because he’s a bad person. But because he doesn’t have the instruments to be in the world any other way. And the camera knows it. It stays with him in the empty spaces, in the diary monologues, in the nights in the cab where the city passes like something foreign and hostile. Scorsese builds Travis’s loneliness not as an emotional state but as a physical condition. Something that occupies space in every frame.
“Loneliness has followed me my whole life. Everywhere. In bars, in cars, on sidewalks, in stores, everywhere. There’s no escape. I’m God’s lonely man.”
Travis borrows a line from Thomas Wolfe as if it were his own, but misreads it. Unlike Wolfe, who stipulates that loneliness is inherent to all men, Travis feels chosen by it. That distortion is not just another character trait. It is the engine of the entire story.
That is what makes the third act work without needing resolution in the conventional sense.
The violence of the final stretch is not a twist. It’s not a narrative surprise calculated for impact. It’s a consequence so direct from everything before it that when it arrives there’s something almost logical about it, in the most disturbing sense of the word. Travis doesn’t make a decision that changes his path. His path leads him there inevitably. Not because violence is the only possible outcome for a character like this in moral terms, but because it’s the only possible outcome for this specific character, built in this specific way, in this specific story. The film couldn’t end any other way because from the beginning it could only go in one direction.
That’s the distinction that matters. It’s not about the story being dark or the character being irredeemable. It’s about internal coherence. About the ending emerging from the logic the story has established, not from the logic the screenwriter needs at the last moment to close it.
What happens after the violence is even harder to build and more revealing about what Schrader understood he was doing. Society interprets Travis’s act as heroic. He receives recognition. Iris returns to her family. On the surface, there is resolution. The narrative elements seem to close. But it’s a false resolution, and Scorsese builds it with that falseness deliberately exposed, without commentary, without underlined irony, with the same coldness with which he has filmed everything before. Nothing that defined the conflict has actually changed. Travis has understood nothing about himself. He hasn’t evolved in any real sense. He hasn’t resolved his relationship with the world or with his own violence. He has only found, momentarily, a way to fit into it. And the last image of the film, that glance in the rearview mirror, suggests the cycle is about to begin again. That Travis will keep driving. That the city will keep being what it is. That nothing has ended.
That’s what makes the film keep generating discomfort decades after it was made. Not because it leaves questions unanswered in the sense of an unresolved plot. But because it shows, with absolute clarity, that the problem remains intact. That the apparent resolution hasn’t touched anything essential. And that integrity, that refusal to offer a way out the story hasn’t earned, is exactly what most screenplays can’t sustain when they reach the end.
Because sustaining it demands something that can’t be improvised in the third act.
It demands having built from the beginning something that determines how it must end. Not as a thematic decision made in the abstract. Not as an intention declared in the treatment. As a structural consequence of each scene, of each character decision, of each moment when the pressure could have eased and didn’t. If Taxi Driver ended with a genuine transformation of Travis, with some kind of emotional opening that suggested possible redemption, it would betray its own logic from the first scene. If it offered a clean catharsis, it would deny everything Schrader put into those pages written in solitude and urgency. The ending works precisely because it doesn’t try to resolve what the story has established cannot be resolved.
And there lies the real problem with the screenplays that never reach that point.
It’s not that the screenwriter doesn’t know how to close. Most of the time they know how to close. The problem is earlier and deeper. They’ve built something that generates one kind of ending and then try to write another because that ending makes them uncomfortable, because it wasn’t what anyone expected, because someone in the process said the protagonist needs a clearer arc. Or worse: they’ve built something that can’t close cleanly because the story demands it, and they’re still looking for the closure because they don’t trust the audience to hold that openness.
In both cases the result is the same. The audience feels something doesn’t add up without being able to identify exactly what. There’s a dissonance between what the story promised and what it delivers. And that dissonance doesn’t live in the last pages. It lives in the moment the story stopped being honest with its own logic.
Where you relieved the pressure when you shouldn’t have. Where you introduced a comic relief scene that broke the accumulation. Where you resolved a tension that had to stay open because not closing it made you anxious. Where you decided it was enough when it still wasn’t. Where you started writing for the ending you wanted instead of the ending you needed, the ending the story was building.
That’s where the problem starts. And that’s where it really has to be corrected.
The audience doesn’t need a perfect ending. They don’t need an ending that satisfies them in the most comfortable sense of the word. They need an inevitable ending. One that, when it arrives, makes everything before it take on a meaning it didn’t have before in such an explicit way. One that cannot be any other way without betraying what is essential. The story.
That inevitability is not built at the end. It is dragged to the conclusion from the beginning, scene by scene, moment by moment, without yielding, without making deals, without relieving what must not be relieved.
If your story doesn’t reach that point, don’t look at the last act.
Look at where you stopped dragging and started inventing.