Type "merge polygons" into a search box with a layer of land parcels open in QGIS and you will be offered three different operations wearing one name. Only one of them is probably yours. Most people who say merge polygons mean: my layer has too many pieces, the pieces that share a value should be one feature, and the internal boundaries between them should disappear. That operation is Dissolve, and it is a small tool with exact behavior that repays exact understanding. This guide teaches Dissolve as the main road, then marks the two turnings it is confused with, so the next time the phrase comes up you take the right one first try.
Home / GIS and Open Map Data for Beginners
Merge Polygons in QGIS: Dissolve, Done Right

Photo: Sam McCool / Pexels
One Layer, Fewer Features: The Dissolve Checklist
Check off each step as you finish it. Your progress stays on this page.
0 of 7 complete
0% done
Dissolve: The Operation Behind the Phrase
Find it at Vector, Geoprocessing Tools, Dissolve, or in the Processing toolbox as the native algorithm native:dissolve. Its contract is precise:
- Input: one vector layer.
- Dissolve fields: zero or more attribute fields. Features are grouped by their combination of values in those fields.
- Output: one feature per group, whose geometry is the union of the group's geometries, with the internal boundaries dissolved away.
The worked example that fixes the idea: a layer of 214 municipal parcels carries a district field with 6 values. Dissolve on district returns 6 features, one per district, each the fused shape of its parcels, the parcel lines inside it gone. The count check, 214 in and 6 out, is the verification habit this guide will keep returning to: dissolve is one of the rare operations whose correctness you can audit by arithmetic before you ever look at the map.
The Multipart Behavior, and the Switch That Controls It
Dissolve has a subtlety that produces most of its support questions. When features with the same value are not adjacent, a peninsula of District A across the river from the rest of District A, the default output keeps them as one multipart feature: a single row in the attribute table whose geometry is several separated polygons. That is usually correct for analysis (the district is one thing) and occasionally wrong for the next step (some tools and formats handle multipart poorly, and a label or centroid lands in unexpected water).
The switch is in the advanced parameters: Keep disjoint features separate. Checked, nonadjacent same-value features come out as separate single-part features, one row each, and the 6-district example might return 9 rows if three districts are split by rivers. Neither setting is the right one in general; the right one is the one your next operation expects, and the count check is how you confirm you got it: predict the row count under each setting before you run, then make the output explain itself.
The Blank Field List: Dissolve Everything Into One
Run Dissolve with no dissolve fields at all and the grouping key is empty, so every feature belongs to the single group and the output is one feature: the union of the entire layer, internal boundaries gone. This is a legitimate and common job, turning a layer of county tiles into the state outline, a set of survey plots into the project boundary, and it is also the accident that befalls people who open the dialog, click Run, and discover their 214 parcels became a blob. The blob is not a bug; it is the blank-field contract, honored exactly. Undo, choose the field, and run again, and let the incident teach the dialog's one lesson: in Dissolve, the field list is the operation.
Dissolve Does Not Clean: Fix Geometries First
The second great support question is the dissolve that produces garbage: slivers, self-intersections, a union with holes where none belong. The cause is upstream. Dissolve performs a union, and union arithmetic on invalid geometry, polygons that cross themselves, double back, or carry duplicate nodes, produces invalid results with total confidence. QGIS ships the remedy as its own operation, and our guide to fixing geometry in QGIS owns it: run the validity check, repair with Fix geometries, and only then dissolve. The professional order is check, fix, dissolve, count. Skipping to step three is how a ten-minute job becomes an afternoon of wondering which tool lied, when none of them did.

The Other Two Merges, Named So You Can Refuse Them
Merge Vector Layers stacks separate layers into one file: parcels.shp plus parcels2.shp becomes one layer with all 428 features, boundaries and rows intact. Different input (many layers), different output (the same features, filed together), different verb. Our guide to merging layers in QGIS owns it, and the one-line separation is worth memorizing: merging layers changes how many files you have; dissolving changes how many features you have.
Merge Selected Features, in the editing toolbar, fuses the features you have selected by hand into one, inside an edit session, asking which feature's attributes survive. It is the surgical instrument for the six polygons you are looking at, where Dissolve is the policy applied to the whole layer by rule. Hand-merging a layer of thousands by selection is a category error; dissolving a six-feature fix by field is another. Choose by scale: a rule for the layer, a hand for the exception.
Attributes After the Merge: What Survives
A dissolved feature carries the attributes of its group key and, for the other fields, values inherited from a representative feature, which is a polite way of saying: do not trust the surviving owner or area values on a dissolved row to describe the group. If the analysis needs sums, areas, or counts per group, compute them deliberately: the field calculator and the statistics tools aggregate by the same field you dissolved on, and pairing the dissolve (for geometry) with an aggregate table (for numbers) is the complete job. Dissolve alone gives you the shapes of the groups; the numbers about the groups are a second, separate step that beginners skip and professionals schedule.
Dissolve Fields in Combination: The Grouping Is the Analysis
The dissolve field list accepts more than one field, and the combination is where the tool stops being a cleanup and becomes an analysis. Dissolving parcels on district alone answers "show me the districts." Dissolving on district and zoning together answers the planner's real question, "show me the residential parts of each district as shapes," and the output count is the count of combinations actually present, not the product of the two lists, a distinction the count check catches when a predicted 30 comes back as 23 because seven combinations simply do not occur on the ground. Two disciplines govern multi-field dissolves. First, order does not matter to the grouping, but completeness does: every field that defines a class in your head must be in the list, because the field you omit is a distinction the output erases, silently, at scale. Second, sparse combinations are information: the combination that returns a single sliver feature is the data pointing at an anomaly, a misclassified parcel, a boundary error, and the dissolved output, read with the count check in hand, is one of the cheapest data audits in GIS. Run the dissolve you need for the map; then read the row counts as a report on the layer itself.
What Dissolve Does to Area and Perimeter
One computational consequence deserves its own warning, because it corrupts results quietly. A dissolved feature's area is the honest area of the fused shape, and its perimeter is the fused shape's true boundary, the internal edges gone. Both are correct, and both break naive comparisons: a district whose parcels summed to 41 kilometers of boundary will report a fraction of that after dissolving, because most of the boundary was internal. Reports that compare pre-dissolve and post-dissolve measurements without noting the operation are comparing a pile of pieces with the assembled object. The rule is the one this guide keeps returning to: dissolve is a change of what a feature is, from parcel to group, and every measurement downstream must be read in the new ontology. Analysts who internalize that stop being surprised by their own outputs, which is most of what experience in this trade consists of.
The Workflow, Compressed
- Check geometry validity; fix what is broken.
- Decide the grouping field, and predict the output row count, including the multipart question.
- Run Dissolve, with Keep disjoint features separate set by the prediction, not by default.
- Verify: row count matches the prediction; spot-check one known group on the map.
- Aggregate the numbers the analysis needs, as their own step.
- Save the result as its own layer, in a GeoPackage if the project is one, and keep the undissolved original untouched beneath it.
Related Guides
If this page answered part of the question, these GIS and open data guides cover the rest of the same ground:
- How to Merge Layers in QGIS: One Layer, Same Geometry, Shared CRS: The sibling operation: stacking separate layers into one file, features untouched.
- Network Analysis in QGIS: Shortest Paths and Service Areas on Your Own Streets: Once the polygons are fused, the next analysis family: routes and service areas on a network.
- Viewshed Analysis in QGIS: What a Point Can Actually See: Another named operation with its own plugin and its own discipline.
- Fix Geometry in QGIS: Find Invalid Polygons and Repair Them to Zero Errors: The mandatory first step before any union arithmetic.
Frequently Asked Questions
What is the difference between dissolve and merge layers in QGIS?
Dissolve works inside one layer: it groups features by an attribute value and removes the boundaries between them, producing fewer features. Merge Vector Layers stacks several layers into one file with every feature and boundary preserved. Dissolve changes the feature count; merging layers changes the file count.
Why did my dissolve turn everything into one polygon?
Because the dissolve field list was empty. With no fields chosen, every feature falls into a single group and the output is one fused feature. Undo, select the field whose values define your groups, and run again.
Why are my dissolved features multipart?
Because features sharing a value are not always adjacent, and Dissolve groups by value, not by contact. The default keeps separated pieces as one multipart feature. Check Keep disjoint features separate in the advanced parameters if you need each piece as its own row.
Why did my dissolve produce slivers and broken shapes?
Invalid input geometry. Dissolve unions what it is given, and self-intersecting or otherwise invalid polygons union badly. Run the geometry validity check and Fix geometries first, then dissolve.
Which attributes does a dissolved feature keep?
The grouping fields' values, and representative values for the rest, which you should not rely on for analysis. If you need totals or counts per group, aggregate them as a separate step using the same field you dissolved on.
Photo credits: Sam McCool (Pexels); Matthias Groeneveld (Pexels). Photos are served from this site; follow the credit to the library page for the original.