DEV Community

BM lm
BM lm

Posted on

font-size units explained: px, rem, em, ch, and when each one wins

Five units can set a font-size in CSS, and each one answers a different question. Using the wrong one is how you get a modal whose text scales but whose padding does not.

px - fixed, predictable, wrong for text

font-size: 16px ignores the user's browser font-size preference. Some users set 20px defaults because they need larger text; px sizing silently overrides that. Use px for borders, shadows, and decorative offsets - never for reading text.

rem - the reading unit

1rem = the root (html) font-size, which respects the user's preference. Body text, headings, and spacing for text containers belong in rem. One line: html { font-size: 100%; } and everything in rem now scales with user settings.

em - the component unit

em is relative to the element's own font-size - which makes it perfect for things that should scale WITH the text: padding around a button whose label is in rem, icon sizing inside a heading. The catch: em compounds in nested elements, which is where the bugs live.

ch - the measure unit

1ch = the width of the "0" glyph. The readable line length is 45-75 characters, so max-width: 70ch gives you correct line length at any font size. This is the single most underused unit in typography CSS.

The winning combination

html { font-size: 100%; }
body { font-size: 1rem; line-height: 1.5; max-width: 70ch; }
button { padding: 0.5em 1em; } /* scales with its label */
Enter fullscreen mode Exit fullscreen mode

When units meet fonts

Typography audits sometimes reveal that a "mystery font" renders differently across breakpoints because em-based sizing compounded unexpectedly. When you inherit such a page, identify what is actually rendering first - a screenshot through an on-device identifier like Typeye names the real typeface locally - then normalize the units. Fix the units, keep the font.

Start with rem for text, em for components, ch for measures. px stays in the decorative layer.

Top comments (0)