Blog · Typography research
Serif vs sans-serif for dyslexic readers - what the research says
"Use a sans-serif font" is the single most repeated piece of dyslexia typography advice on the internet. It appears in NHS guidance, in the British Dyslexia Association's style guide, in university accessibility manuals and in roughly every blog post on the subject. It is also, on closer inspection, more nuanced than its repetition suggests. Here is what the research actually shows, why the sans-serif rule became the default, and the small set of cases where a serif is the better choice.
The short answer
For most dyslexic readers, on a screen, a well-chosen sans-serif is the safer default. The evidence behind this is real but narrower than usually stated: it concerns body text on screens at typical reading sizes, and the gains are mostly about subjective comfort and reduced visual noise, not raw speed.
A serif is not "wrong" for dyslexia. If a reader is fluent and the serif is chosen well (high x-height, low contrast between thick and thin strokes, clear letter disambiguation), serif body text can be just as comfortable. The cases where serifs become a problem are small print, low-resolution screens, and decorative serifs with strong stroke contrast.
What "serif" and "sans-serif" actually mean
A serif is the small finishing stroke at the end of a letter's main strokes - the little feet on the bottom of an "n" in Times New Roman, the bracketed top on a "T" in Garamond. Serif typefaces evolved from carved Roman inscriptions and dominated print for five centuries; they are still standard in books and newspapers.
Sans-serif ("without serif") typefaces strip those finishing strokes and present the letter in its plainer skeleton. Helvetica, Arial, Verdana, Lexend and Atkinson Hyperlegible are all sans-serifs. They feel more modern, are dominant on screens, and tend to look cleaner at small sizes and low resolutions.
Where the "sans-serif for dyslexia" rule comes from
The rule is older than the dyslexia-typography research literature it now claims to rest on. Through the 1990s, accessibility guides for dyslexic readers (and for low-vision readers generally) recommended sans-serif fonts on the reasoning that serifs add visual complexity to each letter, and that complexity costs effort for readers who are already working harder to decode. The reasoning was sound but largely unmeasured.
The British Dyslexia Association's style guide, first published in the 2000s and revised several times since, formalised the recommendation: sans-serif, 12-14 point in print and ideally larger on screen, generous line spacing, no underlining, no italics for whole paragraphs. Most other accessibility guides have copied that line by line.
The empirical research has, by and large, supported the recommendation - but with two important qualifications. First, the gap between a good serif and a good sans-serif is small: in studies that test both fairly (matched x-height, matched size, matched spacing), the differences in objective reading speed are usually within the margin of error. Second, the gap is much larger between a good font and a bad font of either category. A clean serif beats a cluttered sans-serif. A modern, readability-tuned sans-serif (Lexend, Atkinson Hyperlegible) beats Times New Roman. Choosing well within a category matters more than choosing the category.
What the research actually measured
Rello and Baeza-Yates (2013)
One of the most cited studies in this area tested 12 fonts with 48 dyslexic adult readers, measuring both reading speed (via eye-tracking) and subjective preference. The study found that sans-serif and monospaced fonts produced significantly faster reading and shorter fixations than serif and italic fonts. Comic Sans, Helvetica, Arial, Verdana and Computer Modern Unicode performed best; Times, Garamond and Arial Italic performed worst. The italic finding is robust and worth keeping in mind: italic body text is consistently bad for dyslexic readers regardless of whether the underlying font is serif or sans.
Rello and Baeza-Yates (2017) follow-up
A larger follow-up looked at how reading speed varied with font, weight, and other typographic parameters. Sans-serif retained its advantage but the effect size was modest, and once x-height and stroke width were controlled for, the serif/sans-serif distinction explained less of the variance than letter-spacing and line-height did.
The bigger picture
Across the literature, the consistent findings are: italic body text is bad for dyslexia (large effect, well-replicated), tight letter-spacing is bad (medium effect, well-replicated), and "sans-serif beats serif" is true on average but with a small effect that depends heavily on which serif you compare against which sans-serif. For a deeper look at the spacing factors that move the needle more than font choice does, see line spacing and letter spacing for dyslexia.
Why sans-serif tends to win on screens
Several practical factors compound the modest research advantage when you move from print to screens:
Pixel grids. Serifs are small features and small features render badly at low resolutions. On a standard 1080p monitor at typical reading sizes, the serifs on a font like Georgia render as one or two pixels of detail and can become slightly fuzzy or asymmetric. Sans-serifs have fewer small features to lose. On a high-DPI screen (Retina, 4K) this gap closes dramatically.
Font hinting and antialiasing. Modern font rendering does an excellent job, but it still has more to do on a serif glyph than a sans-serif one. Subtle hinting failures show up as inconsistent stroke weights, which is exactly the kind of visual noise that fatigues dyslexic readers faster.
Reading distance and posture. Screens are read at variable distances and angles; books are not. A typeface that is tolerant of changes in apparent size and angle (sans-serif, on average) holds up better in this messy real-world reading environment.
Page width. Web pages tend toward longer line lengths than printed books, and sans-serif typefaces deal with very long lines slightly better because the eye does not need to track across as many small visual features per line.
The cases where a serif is actually better
The "always sans-serif" rule has exceptions. The clearest are:
Long-form print reading. If you are reading a printed book or a high-quality printed magazine, a well-set serif (a book-weight serif like Sabon, Source Serif, or any of the more readable Caslon variants) is genuinely fine for most dyslexic readers, and many actively prefer it. The pixel-grid problem disappears, the serifs help track the eye along the line, and the centuries of optical refinement that went into book typefaces show up.
High-DPI screens at large sizes. On a Retina iPad or a 4K monitor at 18 px or larger, a clean serif (Charter, Source Serif, Iowan Old Style) renders at near-print quality and the rendering disadvantage largely vanishes. Some dyslexic readers find that at this size, the serifs help anchor the eye between words, and they prefer it for long reads.
Habit and confidence. If you have been reading serifs comfortably for years, switching to a sans-serif because the internet says you should is not a clear win. Reading speed and comfort are heavily learned. A reader fluent in a familiar font usually outperforms the same reader with an unfamiliar but technically better font, at least in the short term. Give a new font a real two-week trial before judging it.
Specific "anti-confusion" serifs. A small number of serif typefaces are specifically engineered for high legibility - notably some of the recent "schoolbook" designs aimed at children's reading material. These are not the standard book serifs and they are quite rare; if you encounter one in an educational context, it is not the same animal as Times New Roman.
What about Times New Roman specifically?
Times New Roman deserves its own paragraph because it is the serif most dyslexic readers actually encounter. The honest assessment: it is a competent typeface that has been deployed inappropriately for decades. It was designed in 1931 for narrow newspaper columns at small sizes, optimised for print on absorbent paper, and was never intended as a body font for screens. On screens, at default browser sizes, in long lines, on white backgrounds, Times New Roman performs measurably worse for dyslexic readers than almost any modern sans-serif. This is the comparison most people have in mind when they say "serifs are bad for dyslexia." A fairer comparison (Source Serif vs Atkinson Hyperlegible, both at 18 px on a Retina screen with 1.6 line-height) would not produce the same gap.
Practical decision framework
Use this in roughly the following order of priority:
1. Italic is bad. Whatever else you do, do not read long-form italic body text if you can avoid it. This is the single most robust finding in the literature.
2. Spacing matters more than serif/sans. Generous line-height (1.5-1.7) and slightly increased letter-spacing (0.03-0.06 em) move the needle more than picking a serif vs sans-serif. Get spacing right first.
3. On screens, default to sans-serif. Specifically, use a readability-tuned modern sans-serif - Lexend, Atkinson Hyperlegible, or if you have the specific letter-rotation issue, OpenDyslexic. Plain Arial is acceptable; Verdana is fine for very small UI text.
4. In print at book size, either category is fine. If a printed book is set in a competent book serif at 11-12 point with proper leading, it will work for most readers. If you struggle, the issue is more likely contrast, paper colour, or your environment than the typeface category.
5. Skip the decorative middle ground. Slab serifs (Rockwell, Roboto Slab) and high-contrast serifs (Didot, Bodoni) are the worst of both worlds for dyslexia: visual complexity without the readability tuning. Avoid them as body fonts.
Applying your choice everywhere you read
One practical wrinkle: most websites do not let you change their fonts, and the ones that do (Medium, Substack, some publishers) only offer a handful of preset choices. If you have decided that a particular sans-serif works for you, you need a way to apply it across every site without fighting each one individually.
This is exactly what LexiFont does. It is a Chrome extension that overrides the font on every page with your choice - Lexend, Atkinson Hyperlegible, OpenDyslexic, Comic Neue or your system font. The free tier covers Lexend and the basics; LexiFont Pro unlocks the full set including OpenDyslexic and Atkinson Hyperlegible for a one-time purchase. There is a five-minute test you can run today: switch on the extension, pick one font, read for a week, switch to a different one, read for another week. Subjective comfort over the second paragraph is usually a better guide than any research summary.
What if I have already tried sans-serif and it did not help?
This happens, and it is worth taking seriously. Two common explanations:
The font was not actually a "good" sans-serif. Switching from Times New Roman to Arial is a small step. Switching to Lexend or Atkinson Hyperlegible is a much bigger one because they have been engineered for the specific failure modes (character disambiguation, reading rhythm) that the older sans-serifs ignore. If your "I tried sans-serif" experiment was Arial or Helvetica, try one of the modern dyslexia-tuned ones before concluding sans-serif does not work for you.
The problem is not the font. Font choice is one of perhaps a dozen factors that affect dyslexic reading on screens. The others - font size, line length, line height, page background colour, screen brightness, ambient lighting - sometimes dominate. If a font change has not helped after two weeks, run through those before trying another typeface.
The honest summary
The "use sans-serif" rule is correct on average, modest in effect size, and importantly secondary to spacing, italic-avoidance, and overall page design. It is the right default for dyslexic reading on screens, especially with one of the modern readability-tuned sans-serifs. It is not a hard prohibition on serifs, and switching from a familiar serif to an unfamiliar sans-serif may not feel like an improvement for weeks. Run real trials, prioritise spacing first, and treat font choice as one knob among several rather than the answer.
Get LexiFont Pro - Lexend, OpenDyslexic, Atkinson Hyperlegible and Comic Neue for $14.99 one-time