Caption Reading Speed and Line Length Rules That Work
Play a captioned clip to five people with the sound off and you will usually see the same thing. Two of them finish each line comfortably, two reread the second half, and one gives up and watches the picture. The words were accurate.

Play a captioned clip to five people with the sound off and you will usually see the same thing. Two of them finish each line comfortably, two reread the second half, and one gives up and watches the picture. The words were accurate. The timing and line length were not. Caption reading speed is the part of captioning most editors never measure, and it decides whether anyone actually reads what you typed.
I caption podcasts and short documentaries, and the fixes below come from watching real viewers struggle with my own early files. None of them need special software. They need a number for characters per second, a limit on line length, and the discipline to cut words when the speaker talks faster than people read.
Why Caption Reading Speed Matters More Than Accuracy
A perfectly transcribed caption that flashes past in 1.2 seconds is worse than a slightly condensed one that stays up for 2.5. Viewers do not just read captions. They read, look back at the face or the action, then return to the text. Every glance away costs time, and dense captions punish it.
Professional guidelines put numbers on this. Many broadcasters aim for roughly 15 to 17 characters per second for adult viewers, and lower, around 12 to 13, for children's content. Some streaming services allow up to 20 characters per second for adult programming, but that ceiling is a maximum, not a target. If your editor shows a CPS figure for each event, it is the most useful number on the screen.
Line Length and the Two-Line Rule
Most style guides cap a caption line at 37 to 42 characters, including spaces, and limit each event to two lines. Longer lines force the eye to travel too far across a wide frame and push text toward the edges, where it is harder to track on a phone held upright.
| Setting | Common target | Why |
|---|---|---|
| Characters per line | 32 to 42 | Keeps text near the center of the frame |
| Lines per caption | 2 maximum | Three lines cover too much picture |
| Characters per second | 15 to 17 for adults | Leaves time to glance at the image |
| Minimum duration | About 1 second | Shorter events register as a flicker |
| Maximum duration | About 7 seconds | Longer holds look stuck |
Vertical video for short-form platforms needs tighter limits still. On a 9:16 frame, I keep lines under 28 characters, because the usable width is narrow and platform buttons sit over the right edge.
Where to Break a Line
Line breaks are where good captions separate from auto-generated ones. The rule is simple: break at a natural pause in the grammar, never in the middle of a unit of meaning.
Breaks that read well
- After a comma or full stop
- Before a conjunction such as "and", "but" or "because"
- Before a preposition that starts a new phrase
Breaks that trip readers up
- Between an article and its noun, as in "the" at the end of line one and "microphone" on line two
- Between a first and last name
- Between a verb and its auxiliary, such as "will" and "record"
When you can choose, make the top line shorter than the bottom one. A short top line with a longer base reads like a pyramid and keeps the text visually anchored.
Verbatim Is Not Always the Kind Choice
A lot of accessibility advice says captions should always be fully verbatim. I disagree for most edited content. A guest speaking at 220 words per minute produces more text than a typical viewer can read, and a verbatim file at that pace pushes CPS past 25. The result is accurate captions that nobody can follow.
Condensing is not the same as rewriting. Remove false starts, repeated words and filler like "you know" and "sort of", and keep every piece of meaning and every name. If you have already cleaned filler from the audio edit, your captions will usually follow naturally. The exceptions are legal content, interviews where exact wording matters, and anything where a viewer might quote the speaker. Those stay verbatim, and you slow the pace by extending durations into pauses instead.
Timing Captions to Shot Changes
A caption that hangs two or three frames across a cut is a small thing that viewers notice without knowing why. The text appears to jump, and the eye resets. Most guidelines suggest either ending a caption at the cut or holding it at least a dozen frames past it, never a handful of frames either side.
My routine in the subtitle editor runs in this order:
- Import the transcript and let the tool auto-sync to the waveform.
- Turn on shot-change markers if the editor offers them.
- Snap any caption that ends within six frames of a cut to the cut itself.
- Add a gap of at least two frames between consecutive captions so they register as separate events.
- Sort the list by CPS and fix the twenty worst events first.
On a typical 30 minute podcast episode, that last step takes me about 20 minutes and catches most of the captions viewers would have complained about. Fixing the worst events first matters because they cluster: a fast speaker, a heated exchange, a list read aloud. Solve those passages and the average CPS for the whole file usually drops by two or three points.
Checking Your File Before Export
The last pass is a viewing pass, not a reading pass. Watch the full piece with audio muted at normal speed, on the device most of your audience uses. If you catch yourself pausing, that caption needs work. Then confirm the format will keep your line breaks intact, because some platforms rewrap text and quietly undo your work. Our guide to choosing a caption file format that survives re-uploads covers which formats hold their breaks.
Speaker changes need their own attention too. A new line per speaker, with a dash or a name label, keeps reading speed manageable in conversations. The approach I use for labelling speakers in a video podcast fits the same line limits described here.
FAQ
What is a good characters per second limit for YouTube captions?
Aim for 15 to 17 characters per second for general adult content, and never exceed about 20. For tutorials where viewers also watch the screen closely, staying near 14 gives them room to follow both.
Should captions and subtitles follow the same line length?
Mostly yes. Subtitles for translated dialogue often run slightly shorter because translation can compress text, but the two-line cap and the 37 to 42 character range apply to both in most style guides.
Do auto-generated captions follow these rules?
Rarely. Automatic captions often break lines by character count alone, ignore shot changes and keep every filler word. Treat them as a draft transcript, then apply your own CPS, line length and break rules.
The Practical Takeaway
Pick three numbers and write them on a sticky note: 42 characters per line, two lines per caption, 17 characters per second. Sort every new file by CPS before export and fix the worst events first. You can find more workflows in our captions section, but those three numbers alone will make your next file easier to read than most of what gets uploaded this week.
More in Captions and Subtitles
Captions and Subtitles
Labelling Speakers in Captions for a Video Podcast
A video podcast with three people at one table is the hardest thing I caption, and it is not because of the words.
Captions and Subtitles
Choosing a Caption File Format That Actually Survives Re-Uploads
A creator I worked with lost three hours of careful caption timing in a single export mistake.
Captions and Subtitles
The Caption Mistakes That Make Viewers Stop Watching
Captions get treated as an accessibility checkbox, something added at the end of the editing process to satisfy a platform...