Content Creation

How eye-tracking research, streaming style guides, and accessibility law are shaping the way we size and time subtitles this year. Every subtitle on screen is a small negotiation between two competing needs: the viewer’s eyes have to leave the image, decode a block of text, and return to the picture before anything important happens. Get the timing or the size wrong, and one of two things happens — either the viewer misses the dialogue, or they miss the video. Readability isn’t a style preference. It’s a measurable outcome, and in 2026 it’s backed by more data than ever: eye-tracking studies, platform-published style guides, and accessibility regulations that are now legally enforceable across major markets. This guide breaks down exactly what “readable” means for subtitles today — the font sizes that hold up across devices, the reading-speed limits that separate professional captions from amateur ones, and the layout choices that keep text out of the way of the story. Whether you’re publishing a YouTube tutorial, a vertical Reel, or a broadcast-ready documentary, these numbers are the difference between subtitles people read comfortably and subtitles people give up on. Why Subtitle Readability Is a Science, Not a Guess Reading subtitled video is a fundamentally different cognitive task than reading a book or a webpage. The viewer’s gaze has to split its attention between two sources of information — the moving image and the on-screen text — and constantly decide where to look next. Eye-tracking research on subtitled content consistently shows the same pattern: viewers spend a large share of their viewing time fixated on the subtitle line itself, glancing up at the picture in short bursts. If the text is too small, too dense, or on screen for too short a time, that balance collapses. Viewers either abandon the picture to keep reading, or abandon the subtitle to keep watching — and comprehension drops either way. That’s why the two variables covered in this article — font size and reading speed — sit at the center of subtitle design. They are the two levers most directly tied to whether a viewer can physically finish reading a line before it disappears, and whether they can do it without straining. Everything else (font choice, color, position) supports those two numbers; it doesn’t replace them. Optimal Subtitle Reading Speed: The CPS Standard Reading speed for subtitles is measured in characters per second (CPS) — the total character count of a subtitle line divided by the number of seconds it stays on screen. CPS is preferred over words-per-minute because word length varies wildly between languages, while character count stays a consistent, comparable unit. The formula: CPS = total characters in the cue ÷ display duration in seconds. A 51-character subtitle shown for 3 seconds runs at 17 CPS. Here’s how the major reading-speed benchmarks compare heading into 2026: Audience / Platform Recommended CPS Notes General adult audience 15–17 CPS Widely cited as the comfortable baseline across broadcasters and researchers Netflix (adult content) Up to 17–20 CPS 20 CPS is the outer limit; 17 CPS is the recommended target Netflix (children’s content) Up to 13 CPS Slower pace accounts for developing reading skill BBC Under 15–17 CPS (≈160–180 WPM) Designed around broadcast audiences, including older and less-fluent readers FCC (US broadcast) Up to 18 CPS Regulatory ceiling for captioned broadcast television CJK languages (Chinese, Japanese, Korean) 9–12 CPS Each character carries more information, so lower CPS reads at an equivalent pace to 17–20 CPS in Latin scripts Push past roughly 20 CPS and most adult viewers can no longer finish reading a line before it disappears — they either fall behind or stop reading text entirely and rely on audio and visuals alone. Go much below 12 CPS, and subtitles start to feel sluggish, lingering on screen after the line has already been read. The sweet spot for most 2026 content — social, educational, corporate, and long-form video alike — is 15 to 17 CPS, with a hard ceiling around 20 CPS for fast-paced dialogue. Minimum and Maximum Cue Duration CPS alone doesn’t capture everything. Two additional timing rules matter just as much: Why This Number Is Higher Than It Used To Be Streaming platforms pushed reading-speed limits upward over the last decade. Traditional broadcast subtitling in many regions was built around 10–12 CPS; Netflix’s style guide normalized 17 CPS, with flexibility up to 20. That shift reflects an audience trained on fast-cut, caption-heavy content — but it isn’t universal. Regulators in the UK and EU are actively working to bring streaming captioning in line with traditional broadcast accessibility standards, and research comparing regions shows real friction between fast, streaming-era pacing and audiences accustomed to slower, more traditional subtitling. The practical takeaway for 2026: 17 CPS is a safe, well-supported default. Treat 20 CPS as a ceiling for fast dialogue, not a target, and drop to 12–13 CPS for children’s content, educational material, or any audience that skews toward less-fluent readers. Optimal Subtitle Font Size for 2026 Font size isn’t a fixed pixel number — it’s a ratio. What matters is how large the text appears relative to the frame it sits in and the distance the viewer sits from the screen. A rule of thumb used across broadcast and accessibility guidelines: set subtitle text height to somewhere between 1/20 and 1/10 of the frame height, with a recommended floor of around 44px for HD (1080p) video. From there, the exact number shifts by platform and orientation. Format / Platform Source Resolution Recommended Font Size Horizontal long-form (YouTube, LinkedIn) 1920×1080 24–32 px (≈16–18 pt on desktop-style previews) Vertical short-form (TikTok, Reels, Shorts) 1080×1920 36–48 px, often pushed to 48–60 px for maximum legibility Square (Instagram feed) 1080×1080 28–36 px Broadcast TV delivery 1920×1080 30–36 px, per network spec 4K export (any format) 3840×2160 Double the 1080p value to preserve the same visual size Vertical, mobile-first formats consistently need larger type than horizontal video. There are two reasons: phones are viewed at arm’s length rather than across