diff --git a/CHANGELOG.md b/CHANGELOG.md
index 6fd0f3a3db3d..b4daaa00ee99 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,7 @@ https://github.com/opencv/cvat/issues/750).
 - Mask-RCNN Auto Annotation Script in OpenVINO format
 - Yolo Auto Annotation Script
 - Auto segmentation using Mask_RCNN component (Keras+Tensorflow Mask R-CNN Segmentation)
+- Added MOT CSV format support
 - Ability to dump/load annotations in LabelMe format from UI
 
 ### Changed
diff --git a/README.md b/README.md
index 9bacba82e981..b3d0480acfd0 100644
--- a/README.md
+++ b/README.md
@@ -43,6 +43,7 @@ Format selection is possible after clicking on the Upload annotation / Dump anno
 | [MS COCO Object Detection](http://cocodataset.org/#format-data)                    | X      | X      |
 | PNG mask                                                                           | X      |        |
 | [TFrecord](https://www.tensorflow.org/tutorials/load_data/tf_records)              | X      | X      |
+| [MOT](https://motchallenge.net/)                                                   | X      | X      |
 | [LabelMe](http://labelme.csail.mit.edu/Release3.0)                                 | X      | X      |
 
 ## Links
diff --git a/cvat/apps/annotation/mot.py b/cvat/apps/annotation/mot.py
new file mode 100644
index 000000000000..b7f63d1e79b0
--- /dev/null
+++ b/cvat/apps/annotation/mot.py
@@ -0,0 +1,109 @@
+# SPDX-License-Identifier: MIT
+format_spec = {
+    "name": "MOT",
+    "dumpers": [
+        {
+            "display_name": "{name} {format} {version}",
+            "format": "CSV",
+            "version": "1.0",
+            "handler": "dump"
+        },
+    ],
+    "loaders": [
+        {
+            "display_name": "{name} {format} {version}",
+            "format": "CSV",
+            "version": "1.0",
+            "handler": "load",
+        }
+    ],
+}
+
+
+MOT = [
+    "frame_id",
+    "track_id",
+    "xtl",
+    "ytl",
+    "width",
+    "height",
+    "confidence",
+    "class_id",
+    "visibility"
+]
+
+
+def dump(file_object, annotations):
+    """ Export track shapes in MOT CSV format. Due to limitations of the MOT
+    format, this process only supports rectangular interpolation mode
+    annotations.
+    """
+    import csv
+    import io
+
+    # csv requires a text buffer
+    with io.TextIOWrapper(file_object, encoding="utf-8") as csv_file:
+        writer = csv.DictWriter(csv_file, fieldnames=MOT)
+        for i, track in enumerate(annotations.tracks):
+            for shape in track.shapes:
+                # MOT doesn't support polygons or 'outside' property
+                if shape.type != 'rectangle':
+                    continue
+                writer.writerow({
+                    "frame_id": shape.frame,
+                    "track_id": i,
+                    "xtl": shape.points[0],
+                    "ytl": shape.points[1],
+                    "width": shape.points[2] - shape.points[0],
+                    "height": shape.points[3] - shape.points[1],
+                    "confidence": 1,
+                    "class_id": track.label,
+                    "visibility": 1 - int(shape.occluded)
+                })
+
+
+def load(file_object, annotations):
+    """ Read MOT CSV format and convert objects to annotated tracks.
+    """
+    import csv
+    import io
+    tracks = {}
+    # csv requires a text buffer
+    with io.TextIOWrapper(file_object, encoding="utf-8") as csv_file:
+        reader = csv.DictReader(csv_file, fieldnames=MOT)
+        for row in reader:
+            # create one shape per row
+            xtl = float(row["xtl"])
+            ytl = float(row["ytl"])
+            xbr = xtl + float(row["width"])
+            ybr = ytl + float(row["height"])
+            shape = annotations.TrackedShape(
+                type="rectangle",
+                points=[xtl, ytl, xbr, ybr],
+                occluded=float(row["visibility"]) == 0,
+                outside=False,
+                keyframe=False,
+                z_order=0,
+                frame=int(row["frame_id"]),
+                attributes=[],
+            )
+            # build trajectories as lists of shapes in track dict
+            track_id = int(row["track_id"])
+            if track_id not in tracks:
+                tracks[track_id] = annotations.Track(row["class_id"], track_id, [])
+            tracks[track_id].shapes.append(shape)
+        for track in tracks.values():
+            # Set outside=True for the last shape since MOT has no support
+            # for this flag
+            last = annotations.TrackedShape(
+                type=track.shapes[-1].type,
+                points=track.shapes[-1].points,
+                occluded=track.shapes[-1].occluded,
+                outside=True,
+                keyframe=track.shapes[-1].keyframe,
+                z_order=track.shapes[-1].z_order,
+                frame=track.shapes[-1].frame,
+                attributes=track.shapes[-1].attributes,
+            )
+            track.shapes[-1] = last
+            annotations.add_track(track)
diff --git a/cvat/apps/annotation/settings.py b/cvat/apps/annotation/settings.py
index 1c42bf3b04aa..e1e5f82b42c8 100644
--- a/cvat/apps/annotation/settings.py
+++ b/cvat/apps/annotation/settings.py
@@ -12,5 +12,6 @@
     os.path.join(path_prefix, 'coco.py'),
     os.path.join(path_prefix, 'mask.py'),
     os.path.join(path_prefix, 'tfrecord.py'),
+    os.path.join(path_prefix, 'mot.py'),
     os.path.join(path_prefix, 'labelme.py'),
 )
diff --git a/cvat/apps/engine/tests/test_rest_api.py b/cvat/apps/engine/tests/test_rest_api.py
index ae109bc72723..15bfea7abad8 100644
--- a/cvat/apps/engine/tests/test_rest_api.py
+++ b/cvat/apps/engine/tests/test_rest_api.py
@@ -2662,6 +2662,9 @@ def _get_initial_annotation(annotation_format):
                 annotations["shapes"] = rectangle_shapes_with_attrs + rectangle_shapes_wo_attrs + polygon_shapes_wo_attrs
                 annotations["tracks"] = rectangle_tracks_with_attrs + rectangle_tracks_wo_attrs
 
+            elif annotation_format == "MOT CSV 1.0":
+                annotations["tracks"] = rectangle_tracks_wo_attrs
+
             return annotations
 
         response = self._get_annotation_formats(annotator)