How to Write an Unreliable Narrator

When Wayne Booth coined the term "unreliable narrator" in The Rhetoric of Fiction in 1961, he was naming something that readers had been experiencing for centuries without quite having the language for it. His definition was precise: a narrator is unreliable when their account of events diverges, in ways the reader must recognize, from what the implied author — the sensibility that gives shape to the whole novel — is actually communicating. The unreliable narrator speaks. The novel says something different. The gap between those two things is where meaning lives.

What Booth's definition excludes is important: every narrator is limited. Every first-person voice filters events through one consciousness, omits what it doesn't know, and renders the world in a particular emotional register. Limitation is simply the condition of having a point of view. Unreliability is something more specific — it occurs when the narrator's account is compromised in a way that creates significance, when the reader is meant to understand something the narrator cannot or will not admit. The distinction matters enormously, because conflating the two leads writers toward a technique they barely control and readers toward the exhausting suspicion that no narrator can be trusted about anything at all.

"A novelist is someone who has amnesia and sits down to remember what happened."

— Marilynne Robinson, Paris Review interview, 2008

Robinson's remark illuminates the central problem of first-person fiction: every narrator is, to some degree, reconstructing rather than simply reporting. The unreliable narrator takes this further — the reconstruction is distorted not just by the ordinary forgetting and shaping that all memory entails, but by something more consequential: vanity, trauma, delusion, or the active desire to conceal. Understanding which distortion your narrator suffers from, and what that distortion reveals, is the first and most important decision a writer makes when building an unreliable voice.

The four varieties — and the work each one does

There is a tendency in writing advice to treat unreliable narrators as a single category, as though any narrator who lies or distorts belongs to the same species. In practice, the four main types of unreliability produce very different effects, and conflating them produces very different failures.

The naïve narrator — a child, an outsider, someone with a developmental or cognitive difference — lacks the conceptual equipment to understand what they are reporting. Jack in Emma Donoghue's Room, five years old and raised entirely inside a garden shed, describes his captivity in the only terms he knows: the outside world, glimpsed through a skylight, is simply "outer space." His unreliability is not deception but innocence, and the gap between what he sees and what the reader understands produces a horror far more devastating than any authorial intrusion could achieve. The naïve narrator demands that the writer maintain two registers simultaneously: the character's authentic limited perception, and the contextual information that allows the reader to understand what the character cannot.

The self-deceptive narrator is arguably the richest type because the distortion is itself the psychological portrait. Kazuo Ishiguro's Stevens in The Remains of the Day never lies in any simple sense — he is scrupulous, even exhausting, about the precision of his recollections — but every careful qualification, every rerouting of conversation away from feeling, reveals a man who has organized his entire inner life around the avoidance of what he actually experienced. The reader slowly understands that Stevens's famous "dignity" is a defense mechanism so total it has consumed him. Nothing in the novel announces this. It accumulates, quietly, in the distance between what Stevens reports and what the reader comes to feel about his life.

The deliberately deceiving narrator — Humbert Humbert in Lolita, Amy Dunne in Gone Girl — presents the most familiar conception of unreliability, and also the most technically demanding. When a narrator is consciously constructing a self-serving account, the writer must maintain two levels of rhetoric simultaneously: the seductive surface of the narrator's own performance, and the substrate of evidence that allows the reader to resist it. Nabokov's Humbert is a virtuoso case because the novel never asks the reader to simply disbelieve him — it asks them to notice what his extraordinary eloquence is doing, and to feel the full weight of what he cannot stop himself from revealing between the lines.

The mentally compromised narrator — trauma, psychosis, grief so profound it reorganizes perception — occupies a more ethically complex position. The reader understands that the account is distorted, but by a force the narrator is not fully in control of. The challenge here is avoiding both the reduction of the narrator to a puzzle to be solved and the romanticization of disorder as literary texture. The distortion must do genuine narrative work, not merely signal that something unusual is happening.

Building trust before breaking it

The single most common failure in fiction that announces itself as unreliable is the narrator who never establishes credibility to begin with. If the reader enters with suspicion already fully activated — scanning every sentence for the lie — the technique collapses into a guessing game, and the genuine emotional stakes that make unreliability powerful never materialize.

What makes Humbert Humbert so disturbing is not that readers see through him immediately. It is that, on first reading, many do not. His rhetoric is brilliant, his suffering (as he presents it) seems genuine, and the extraordinary prose seduces the reader into a kind of provisional trust before the full evidence accumulates. Ishiguro's Stevens works similarly: his precision, his modesty, his evident care for his role give every appearance of reliability. The self-deception only becomes visible slowly, retrospectively, as the reader notices what Stevens consistently declines to feel.

In practical terms, this means the opening of an unreliable narrator's story must do two things at once. It must establish a plausible, even appealing, narrative voice — one that earns the reader's initial confidence — while quietly laying the groundwork for the recognition that will come later. This is not simply a matter of planting "clues." It requires a coherent psychology: the narrator must have reasons for seeing the world as they do, and those reasons must be legible even when the reader does not yet understand what they are reasons for.

Planting the inconsistency — the reader's double reading

The mechanism by which an unreliable narrator's account is exposed to the reader operates primarily through what might be called the double reading: the text that functions on a surface level for the narrator's own purposes, and the same text that reveals, to the alert reader, what the narrator is suppressing or distorting.

Secondary characters are the most reliable instrument here. A narrator who consistently misreads other people's responses — who reports someone's cold silence as "thoughtful reserve," someone's obvious distress as "sensitive temperament" — allows the reader to see around the narrator's interpretive framework without the novel ever abandoning that framework. In The Remains of the Day, Miss Kenton's responses to Stevens are reported faithfully and then immediately mis-understood. The reporting is accurate. The understanding is not.

Structural irony — events the narrator describes without grasping their full implication — performs similar work. Ian McEwan's Atonement builds its entire moral architecture around a moment of misinterpretation: Briony Tallis sees something and understands it wrong, and the novel, told substantially from her perspective, asks the reader to hold that wrongness in mind across hundreds of pages. The inconsistency here is not a logical contradiction but an epistemic one — the narrator's certainty about what she saw is real and devastating, and recognizing that certainty as a form of distortion is what the book is about.

What a writer should avoid is what might be called the telegraph error: inconsistencies so obvious and so early that the reader's suspicion is activated before any trust has formed. The double reading works only if there is genuinely something to double: a surface that holds together, against which the contradictions press with increasing weight.

What the reader should understand that the narrator cannot

Unreliability fails — produces confusion rather than revelation — when the writer has not been specific about what, exactly, the reader is supposed to understand. The gap between narrator and truth must point somewhere. It is not enough to create a narrator who is vague or inconsistent or occasionally wrong. The distortion must illuminate something: a psychological pattern, a moral evasion, a capacity for self-deception that tells us something real about how human beings protect themselves from what they cannot bear to know.

Briony Tallis, in her child's certainty, reveals how moral authority can be constructed from misunderstanding — and how the imagination, which is also the source of fiction itself, can become an instrument of harm. Humbert Humbert, beneath the pyrotechnics of his rhetoric, reveals the totalizing structure of predatory self-justification. Stevens, in every deflection and qualification, maps the shape of a life organized around the avoidance of feeling.

In each case, the reader's superior understanding is not a position of simple judgment. The technique works because it implicates the reader too — in the initial trust that was misplaced, in the aesthetic pleasure taken in a voice that was doing something unconscionable, in the retrospective recognition that the clues were always there. The most powerful unreliable narrators leave readers uncertain not just about the facts of the story but about their own readerly behavior. That discomfort is the point.

The question to ask before you commit

Before writing an unreliable narrator, there is one question worth sitting with longer than any of the technical ones: what does this narrator's unreliability reveal? Not about the plot — about the narrator's inner life, about the human pattern their self-deception or delusion or deliberate falsification enacts. If the answer is "it creates an interesting twist" or "it makes the story more surprising," the technique is almost certainly being used as a device rather than as a structural necessity. Tricks work once. Revelation works indefinitely.

The other question is quieter but equally important: is the distortion legible? The reader does not need to decode the truth immediately — in fact, withholding should be one of the technique's pleasures — but the evidence for a different reading must genuinely be present in the text. On re-reading, the unreliable narrator should feel inevitable: every apparent inconsistency resolving into a coherent psychological portrait, every gap between what was said and what was true revealing not confusion but design. That retroactive coherence, the sense that the novel knew exactly what it was doing even when the reader did not, is the mark of the technique used with complete control.

Point of view is always a commitment. The unreliable narrator asks that commitment to carry an additional weight: the reader must believe in the narrator enough to follow them, and understand the narrator clearly enough to see past them. Holding both of those things at once — trust and suspicion, sympathy and recognition — is the peculiar, demanding, irreplaceable experience that only this technique can produce.