What Is the Difference Between Kerning and Tracking and When Would You Use Each

Kerning adjusts the space between one specific pair of letters, while tracking adjusts the spacing uniformly across an entire range of text. Kerning is a micro-level correction applied pair by pair; tracking is a macro-level adjustment applied to whole words, lines, or blocks. You use kerning to fix optical collisions like "AV" or "To," and tracking to open up dense small text or tighten display headlines. The two operate at different scales and solve different problems, which is why professional typography uses both, often on the same page.

The mechanism: pair correction versus uniform distribution

Every glyph in a digital font sits inside an invisible box called the em square, and the space to its left and right is its sidebearing. If you simply stacked those boxes side by side, letterfit would look uneven, because letter shapes have wildly different profiles. A capital "A" is a wedge with empty space at its lower right; a "V" is a wedge with empty space at its upper right. Placed together, those two voids stack into a visible gap.

Kerning data fixes this. Font designers embed kerning pairs, which are exceptions that pull specific letter combinations closer (negative kern) or push them apart (positive kern). The number is expressed in font units, and a typical unit is 1/1000 of an em. A pair like "AV" in a well-built text face might carry a kerning value around -40 to -80 units. At 12 points, that is roughly half a point to a full point of correction, which is small in absolute terms but enormous perceptually.

Tracking works differently. It adds or subtracts a fixed amount of space to every inter-character gap in a selected range, ignoring pair relationships. In design software, tracking is usually expressed in the same 1/1000-em units, so a tracking value of +20 adds 20 units to every gap. A value of -10 removes 10 units from every gap. The adjustment is uniform, mechanical, and blind to shape.

Heres the plain-language version. Kerning asks, "Does this specific pair need help?" Tracking asks, "Does this whole run of text need to breathe or tighten?" One is a scalpel, the other a dial.

Quantified context: how much adjustment is normal

The numbers matter because they tell you when an adjustment is a correction and when it is a redesign.

Kerning values in professionally built fonts typically fall between -100 and +50 units for problematic pairs. Most pairs need zero. This is why automatic kerning produces visible results on only a small fraction of letter combinations. In a typical sentence of 60 characters, you might find 3 to 8 pairs that genuinely benefit from kerning. The rest are already correct.

Tracking values follow a different logic, and the ranges are much wider. Display type set at 48 points or larger often benefits from negative tracking in the -10 to -30 range, because large letters look loosely spaced at default settings. Body text at 9 to 11 points rarely needs tracking changes at all. Small text, meaning anything below about 8 points, often needs positive tracking in the +20 to +50 range to stay legible. All-caps settings commonly take +50 to +150 tracking because capitals have flat sides that resist tight fitting.

The threshold that matters most is size. Below roughly 8 points, positive tracking improves legibility. Above roughly 24 points, negative tracking improves cohesion. Between those, leave tracking alone unless the typeface itself demands otherwise. This is not a law, but it is a reliable starting framework that matches how most professional typefaces are drawn.

The counterintuitive detail: tracking can silently break kerning. When you apply tracking to a range, many layout engines distribute space uniformly and the kerning pairs inside that range may be scaled, overridden, or ignored. In some software the kerning still applies on top of the tracking; in others it does not. The visible result is that a tightly kerned "AV" can open back up when you add tracking to a headline. Professionals check the problem pairs again after any tracking change.

Kerning versus tracking compared

Edge cases and where the models break

Several situations blur the line between the two adjustments.

Justified text is the clearest case. When a layout engine justifies a paragraph, it distributes extra space across word spaces and, depending on the engine, may also alter inter-character spacing. This means tracking is being applied automatically and invisibly, sometimes unevenly across lines. A line with few words may receive aggressive positive tracking, producing visible looseness. This is why justified text in narrow columns often looks worse than ragged-right text; the tracking is doing work the designer never asked for.

Ligatures interact with kerning in ways that surprise people. An "fi" ligature is a single glyph, so there is no inter-character gap to kern inside it. If you apply tracking to a run containing ligatures, the ligature stays rigid while surrounding gaps expand, which can make the ligature look cramped relative to its neighbors. Some software disables ligatures automatically when tracking exceeds a threshold, typically around +10 to +20 units, because the joined forms begin to look wrong.

Letter-spacing in web CSS is the tracking equivalent, and it accepts values in em units. A value of 0.05em adds 5 percent of the font size to each gap. On a 16-pixel font, that is 0.8 pixels per gap, which is subtle. On a 64-pixel headline, it is 3.2 pixels per gap, which is not. The same CSS value produces wildly different visual results at different sizes, which is why responsive typography needs size-specific tracking values rather than a single global rule.

Red-flag warnings worth remembering:

Applying positive tracking to lowercase body text below 10 points usually makes reading slower, not faster, because word shapes fragment. Applying negative tracking to small text closes counters and can make letters like "e" and "a" fill in. Relying on auto-kerning alone for a logo or wordmark is a common failure; logotypes almost always need manual pair work. Setting tracking in pixels instead of relative units breaks when the type scales. Assuming your design software and your browser apply tracking identically will cost you an afternoon of debugging.

Misconceptions that persist

The most common misconception is that kerning and tracking are interchangeable names for "letter spacing." They are not. Letter spacing is the general category; kerning and tracking are two mechanisms inside it, operating at different scopes.

A second misconception is that kerning is obsolete because modern fonts ship with kerning built in. Most professionally made fonts do include kerning tables, but coverage varies enormously. A free font may include a few hundred pairs; a commercial text face may include several thousand. Automatic kerning handles the common pairs and misses the unusual ones, especially in display settings where the letters are large and the errors are obvious.

A third misconception is that tracking should be used to fix bad kerning. This is backwards. If a headline has one bad pair, adding tracking to the whole line opens every gap to hide one problem. The correct move is to kern the problem pair and leave the rest alone.

A fourth, and the professional caveat worth stating plainly: optical kerning and metric kerning are not the same thing. Metric kerning uses the values embedded in the font. Optical kerning uses an algorithm that analyzes glyph shapes and computes spacing from them. Optical kerning often beats metric kerning for display sizes and mixed-font settings, but it can produce inconsistent results across a long document. Professionals choose deliberately rather than accepting whichever default the software applies.

Practical actions with thresholds

Here is a working procedure that holds up across print and screen.

Set tracking first, by size band. For text below 8 points, start at +20 to +40 units. For 9 to 12 points, start at 0. For 13 to 24 points, start at 0 to -5. For anything above 24 points, start at -10 to -25. Adjust from there based on the specific typeface, because typefaces with tight default fitting need less negative tracking than loosely fitted ones.

Kern second, and only the pairs that need it. Scan for the classic offenders: AV, AW, AY, AT, FA, LT, LV, LW, LY, PA, Ta, Te, To, Tr, Tu, Ty, Va, Ve, Vo, Wa, We, Wo, Ya, Ye, Yo. These are the pairs where the diagonal or open side of one letter meets the flat or open side of another. In most text, you will find problems in fewer than 10 percent of these pairs.

Check the interaction. After you set tracking, re-examine the problem pairs. If your software applies kerning on top of tracking, the corrections may still hold; if it does not, you will see the gaps reopen. Test the actual output, not the design file.

Leave body text alone unless it is genuinely too tight or too loose. Most well-drawn text faces at 10 to 12 points need neither adjustment. Tracking body text is a fix for specific problems, not a default step.

For web work, set letter-spacing in em units, and define it per size band in your CSS rather than globally. Test at the actual rendered sizes on real devices, because browser tracking behavior varies more than print.

Limitations and what this cannot tell you

Kerning and tracking guidance is typeface-dependent in ways no universal rule captures. A grotesque with uniform stroke widths tolerates tighter tracking than a humanist face with varied strokes. A script typeface has entirely different fitting logic and may resist both adjustments. The thresholds above are starting points calibrated to common text faces, not guarantees.

Software behavior also varies. Adobe applications, Figma, and browser rendering engines each handle the interaction between kerning and tracking differently, and some engines apply kerning only above a certain size by default. The only reliable test is your actual output pipeline. Numbers in this article describe typical values from professionally built fonts and standard layout engines; they are not measurements of any specific font you are using.

Finally, legibility research does not support a single optimal tracking value. Reading performance depends on typeface, size, contrast, viewing distance, and reader familiarity. The practical guidance here reflects what working typographers converge on, which is a useful proxy but not a controlled experimental result.