Home / GIS and Open Map Data for Beginners

QGIS Symbology: Single, Categorized, and Graduated Styles, and the Classification Choice

Top view of a person working at a desk with two monitors

Photo: Faizur Rehman / Pexels

A layer straight from a file is drawn in one flat color, and it tells you almost nothing. Symbology is the set of choices that turns geometry into information: this layer in one considered symbol, those districts each in their own color, these towns sized by population. In QGIS those choices live in the Symbology tab of Layer Properties, organized by renderers, and the beginner's set has three: single symbol, categorized, and graduated. This guide teaches the three, then slows down for the decision inside the third that actually shapes what your map claims: how a range of numbers gets cut into classes. Advanced renderers exist beyond this article's scope, and it hands them off honestly at the end.

Interactive tool

Style With Intent: The Symbology Checklist

Check off each step as you finish it. Your progress stays on this page.

0 of 7 complete

0% done

Single Symbol: One Layer, One Statement

The Single Symbol renderer draws every feature identically, and it is the right choice more often than beginners think: reference layers, boundaries, the road network under your data, anything whose job is context rather than argument. Do not mistake it for the absence of design. The one symbol still wants a considered color, a width or size that survives the map's final scale, and a transparency that lets it sit under the layers that matter. A map whose background layers are styled with restraint reads as professional before its data layers do anything clever. Set the single symbol with the same care you would give a headline renderer, then leave it alone; its virtue is that it never competes.

Categorized: A Symbol per Kind

The Categorized renderer answers "what kind is each feature." Pick a field, land use, road class, status, and QGIS lists its distinct values, assigning each a symbol: a color for polygons, a line style for roads, a marker for points. You then edit the assignments like an editor, not a spectator: merge the values that mean the same thing for your reader, remove the ones that should not be drawn, choose colors a reader can tell apart at a glance and still name in the legend. Two disciplines keep categorized maps honest. First, the number of categories: past about a dozen, readers stop distinguishing colors and start guessing, so group the long tail into an explicit Other rather than letting fifteen similar hues pretend to be information. Second, color meaning: use hue differences for genuinely different kinds, and resist implying an order (light to dark) among categories that have none; a false gradient is a false claim.

A man working at dual monitors in a home office
A man working at dual monitors in a home office

Graduated: Symbols Scaled by a Number

The Graduated renderer answers "how much," mapping a numeric field onto a ramp of colors or a range of sizes: districts shaded by population density, wells sized by depth, roads widened by traffic. You choose the field, the number of classes, the color ramp or size range, and, the subject of the rest of this article, the classification method that decides where one class ends and the next begins. The map's legend then shows the class ranges, and readers will believe them. That belief is your responsibility: the method is a statistical decision with rhetorical consequences, and QGIS offers five standard ones, each defensible, each telling a different story with the same data.

The Five Classification Methods, Honestly Compared

MethodHow it cuts the rangeStory it tellsWatch out for
Equal IntervalEqual-width slices of the value rangeSimple, comparable across mapsEmpty classes when data clusters
Quantile (Equal Count)Same number of features per classEvery class looks equally importantHides how extreme the extremes are
Natural Breaks (Jenks)Minimizes variance within classes, maximizes it betweenThe data's own clustersBreaks are hard to explain or reproduce by hand
Pretty BreaksRounded, human-friendly boundariesClean legends, easy readingClasses may be unevenly populated
Standard DeviationClasses built around the mean in deviation stepsDistance from average, above and belowOnly meaningful when the mean is meaningful

Equal Interval is the transparent default: if values run 0 to 100 and you ask for five classes, the cuts fall at 20, 40, 60, 80. Its honesty is comparability: two maps with the same range and interval can be read against each other. Its failure is emptiness: when most values huddle between 10 and 30, three of your five classes hold nothing and the map shows one color with a long legend. Quantile guarantees full classes by counting features instead of measuring range, which flatters clustered data into looking evenly spread; the top class may span from 31 to 100 while the bottom spans 0 to 8, and the legend must be read, not glanced, to catch it. Natural Breaks (Jenks) lets the data choose: the algorithm places cuts where values naturally clump, so classes correspond to real groups in the data. It usually produces the most truthful single map of a skewed dataset, and the least explainable legend. Pretty Breaks rounds the boundaries to numbers a human would choose (0, 25, 50, 100), sacrificing statistical neatness for legends people actually read. Standard Deviation reframes the question entirely: classes are steps away from the mean, and the natural styling is a diverging ramp, below average against above, which is the right grammar for anomalies and the wrong one for totals.

The Histogram: Look Before You Classify

QGIS puts a histogram of your field's values inside the graduated panel, and it is the most underused honest tool in the dialog. Look at it before choosing a method. A bell-shaped spread invites Standard Deviation or Equal Interval. A long right tail, a few huge values and many small ones, warns that Equal Interval will empty its top classes and Quantile will hide the tail's drama; Natural Breaks or Pretty Breaks will serve better. Two humps in the histogram are the data telling you where a break belongs, whichever method you choose. Thirty seconds with the histogram converts classification from a superstition into a reading, and the map that results can be defended sentence by sentence, which is the standard a published map should meet. Our print layout and inset map guides take the styled layer from here to paper.

Symbol Levels, Ramps, and the Rest of the Panel

Three supporting features deserve a sentence each. Color ramps are the graduated renderer's palette: sequential ramps (light to dark of one hue) for amounts, diverging ramps (two hues meeting in the middle) for deviations from a center; QGIS ships a library and the ColorBrewer-derived sets are the reliable starters. Symbol levels control drawing order within a layer, the classic use being road casings: draw the wide dark outline of every road first, then the narrower bright fill, by assigning levels so the renderer paints in passes rather than feature by feature. Size and rotation by expression, the data-defined overrides beside most symbol properties, are the bridge from the beginner renderers to the expression system, letting one field steer a property the renderer does not directly offer. Each is a refinement; none rescues a map whose classification is wrong.

Where This Article Stops: The Advanced Renderers

Two renderers sit beyond this guide's scope because each has its own article. The rule-based renderer, which styles features by filter expressions in a decision tree, belongs to our rule-based symbology guide; reach for it when categories and numbers both fail and the styling logic is genuinely conditional. The heatmap renderer, which draws density as a continuous surface, belongs to our heatmap guide; reach for it when the question is where points concentrate rather than what each point is. Knowing which renderer family a question belongs to is most of symbology; the rest is the panel, the histogram, and the willingness to defend your classes.

Saving the Style: QML Files and Style Management

A symbology you have tuned deserves to outlive the layer it was tuned on. QGIS saves a layer's complete styling, renderer, classes, labels, and all, to a QML style file, and loads it onto any other layer with the same fields; the style menu at the bottom of the Symbology panel and the layer's right-click Styles options both speak this language, and copy-style and paste-style move a look between layers inside a session without files at all. Three habits make style files earn their keep. Save the style with the data it styles, in the same folder, named for the layer, so a project reopened in two years can be dressed as it was. Keep a small house library of your standard styles (the boundary style, the survey-points style, the publication style for each data family) so new layers start from the house look rather than from QGIS's cheerful random color. And when a style must travel to other software, know that QGIS also exports SLD, the OGC styled layer descriptor, imperfectly but usefully; the classification breaks and colors survive the trip better than exotic symbols do. A style is an argument about how data should be read, written in a file; treat it with the same care as the data, and your maps will look like one mind made them, because one did.

One Dataset, Three Maps: The Classification Lesson in Practice

Make the lesson concrete with one layer, average income by district, values clustered between modest and comfortable with two wealthy outliers. Under Equal Interval, the two outliers stretch the range so far that every ordinary district lands in the bottom class: the map shows a uniform wash and two specks, and its true statement is "almost everywhere is similar," which the legend's five grand classes disguise. Under Quantile, the districts divide into five equal-count groups and the map blooms into a full ramp, an attractive picture whose top class quietly spans from comfortable to ten times comfortable; its statement is about rank, not amount, and a reader who misses that is misled by beauty. Under Natural Breaks, the cuts find the real structure: the modest cluster, the comfortable cluster, and the outliers standing alone in a class of their own, which is the story the data actually contains. None of the three maps lies in its numbers; each lies or tells the truth in its emphasis. The renderer choice is therefore an editorial act with statistical clothing, and the defense of a published map is never "QGIS chose it." QGIS offered; you chose. Choose like an editor.

Related Guides

If this page answered part of the question, these GIS guides cover the rest of the same ground:

Frequently Asked Questions

Where do I change how a layer is drawn in QGIS?

In the Symbology tab of the layer's Properties. Choose a renderer at the top: Single Symbol for one style for all features, Categorized for a style per distinct value of a field, or Graduated for styles scaled by a numeric field.

What is the difference between categorized and graduated symbology?

Categorized styles by kind: each distinct value of a field gets its own symbol, with no order implied. Graduated styles by amount: a numeric field is cut into class ranges, and symbols progress along a color ramp or size range to show more and less.

Which classification method should I use?

Look at the histogram first. Use Equal Interval for comparable, evenly spread data; Quantile when every class must be populated; Natural Breaks (Jenks) when the data clusters and the map stands alone; Pretty Breaks when the legend must be readable at a glance; Standard Deviation when the story is distance from the average.

Why does my graduated map show only one color?

Usually an Equal Interval classification over clustered data: most features fall into one or two classes and the rest are empty. Check the histogram, and switch to Quantile or Natural Breaks, or set class boundaries manually where the data actually lives.

When should I use rule-based symbology instead?

When the styling logic is conditional in ways categories and classes cannot express, for example different symbols by combinations of fields, or exceptions stacked on a general rule. That renderer has its own guide on this site; the beginner renderers in this article cover most maps.

Photo credits: Faizur Rehman (Pexels); Lisa Fotios (Pexels). Photos are served from this site; follow the credit to the library page for the original.