diff --git a/bin/merge_sdata.py b/bin/merge_sdata.py
new file mode 100755
index 0000000..49168bd
--- /dev/null
+++ b/bin/merge_sdata.py
@@ -0,0 +1,32 @@
+#!/usr/bin/env python
+
+import argparse
+import spatialdata
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Merge SpatialData objects")
+    parser.add_argument("files", nargs="+", help="List of SpatialData files to merge")
+    parser.add_argument("output", help="Output file name")
+    args = parser.parse_args()
+
+    # Read all zarr SpatialData folders
+    sdatas = []
+    for file in args.files:
+        sdata = spatialdata.read_zarr(file)
+        sdatas.append(sdata)
+
+    # Merge the data
+    output_sdata = spatialdata.concatenate(
+        sdatas,
+        region_key=None,
+        instance_key=None,
+        concatenate_tables=False,
+        obs_names_make_unique=True,
+        modify_tables_inplace=False,
+    )
+
+    # Save the concatenated data
+    output_sdata.write(
+        args.output,
+        overwrite=True
+    )
diff --git a/modules/local/merge_sdata.nf b/modules/local/merge_sdata.nf
new file mode 100644
index 0000000..fb4f756
--- /dev/null
+++ b/modules/local/merge_sdata.nf
@@ -0,0 +1,38 @@
+//
+// Merge per-sample SpatialData into a single SpatialData
+//
+process MERGE_SDATA {
+
+    label 'process_low'
+    container "docker.io/erikfas/spatialvi"
+
+    input:
+    path(sdata, stageAs: "?/*")
+
+    output:
+    path("aggregated-sdata.zarr"), emit: sdata
+    path("versions.yml")         , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    if (workflow.profile.tokenize(',').intersect(['conda', 'mamba']).size() >= 1) {
+        exit 1, "The MERGE_SDATA module does not support Conda/Mamba, please use Docker / Singularity / Podman instead."
+    }
+    """
+    # Set environment variables
+    export XDG_CACHE_HOME="./.xdg_cache_home"
+    export XDG_DATA_HOME="./.xdg_data_home"
+
+    # Execute script
+    merge_sdata.py \\
+        ${sdata} \\
+        aggregated-sdata.zarr
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        spatialdata_io: \$(python -c "import spatialdata_io; print(spatialdata_io.__version__)")
+    END_VERSIONS
+    """
+}
diff --git a/subworkflows/local/aggregation.nf b/subworkflows/local/aggregation.nf
new file mode 100644
index 0000000..4a1ebf3
--- /dev/null
+++ b/subworkflows/local/aggregation.nf
@@ -0,0 +1,34 @@
+//
+// Subworkflow for aggregation of sample data
+//
+
+include { MERGE_SDATA } from '../../modules/local/merge_sdata'
+
+workflow AGGREGATION {
+
+    take:
+    ch_sdata // Channel: [ meta, zarr ]
+
+    main:
+
+    ch_versions = Channel.empty()
+
+    //
+    // MODULE: Merge per-sample SpatialData objects into one
+    //
+    ch_sdata_files = ch_sdata
+        | map {
+            meta, zarr ->
+            return [zarr]
+        }
+    MERGE_SDATA (
+        ch_sdata_files.collect()
+    )
+    ch_versions = ch_versions.mix(MERGE_SDATA.out.versions)
+    ch_merged_sdata = MERGE_SDATA.out.sdata
+
+    emit:
+    merged_sdata = ch_merged_sdata // channel: [ aggregated-sdata.zarr ]
+    versions     = ch_versions     // channel: [ versions.yml ]
+
+}
diff --git a/workflows/spatialvi.nf b/workflows/spatialvi.nf
index b51044c..0f380e2 100644
--- a/workflows/spatialvi.nf
+++ b/workflows/spatialvi.nf
@@ -10,6 +10,7 @@ include { MULTIQC                } from '../modules/nf-core/multiqc/main'
 include { INPUT_CHECK            } from '../subworkflows/local/input_check'
 include { SPACERANGER            } from '../subworkflows/local/spaceranger'
 include { DOWNSTREAM             } from '../subworkflows/local/downstream'
+include { AGGREGATION            } from '../subworkflows/local/aggregation'
 include { paramsSummaryMultiqc   } from '../subworkflows/nf-core/utils_nfcore_pipeline'
 include { paramsSummaryMap       } from 'plugin/nf-schema'
 include { softwareVersionsToYAML } from '../subworkflows/nf-core/utils_nfcore_pipeline'
@@ -83,6 +84,14 @@ workflow SPATIALVI {
     )
     ch_versions = ch_versions.mix(DOWNSTREAM.out.versions)
 
+    //
+    // SUBWORKFLOW: Sample aggregation (optional)
+    //
+    AGGREGATION (
+        DOWNSTREAM.out.svg_sdata
+    )
+    ch_versions = ch_versions.mix(AGGREGATION.out.versions)
+
     //
     // Collate and save software versions
     //