diff --git a/README.md b/README.md
index 9d0ba5555..d426fa068 100644
--- a/README.md
+++ b/README.md
@@ -36,15 +36,15 @@ The four-pane view consists of 3 orthogonal cross-sectional views as well as a 3
 
 Neuroglancer itself is purely a client-side program, but it depends on data being accessible via HTTP in a suitable format. It is designed to easily support many different data sources, and there is existing support for the following data APIs/formats:
 
-- [Neuroglancer precomputed format](src/neuroglancer/datasource/precomputed)
-- [N5](src/neuroglancer/datasource/n5)
-- [Zarr v2/v3](src/neuroglancer/datasource/zarr)
+- [Neuroglancer precomputed format](src/datasource/precomputed)
+- [N5](src/datasource/n5)
+- [Zarr v2/v3](src/datasource/zarr)
 - [Python in-memory volumes](python/README.md) (with automatic mesh generation)
 - BOSS <https://bossdb.org/>
 - DVID <https://github.com/janelia-flyem/dvid>
 - Render <https://github.com/saalfeldlab/render>
 - Single NIfTI files <https://www.nitrc.org/projects/nifti>
-- [Deep Zoom images](src/neuroglancer/datasource/deepzoom)
+- [Deep Zoom images](src/datasource/deepzoom)
 
 # Supported browsers
 
@@ -55,7 +55,7 @@ Neuroglancer itself is purely a client-side program, but it depends on data bein
 # Keyboard and mouse bindings
 
 For the complete set of bindings, see
-[src/neuroglancer/ui/default_input_event_bindings.ts](src/neuroglancer/ui/default_input_event_bindings.ts),
+[src/ui/default_input_event_bindings.ts](src/ui/default_input_event_bindings.ts),
 or within Neuroglancer, press `h` or click on the button labeled `?` in the upper right corner.
 
 - Click on a layer name to toggle its visibility.
@@ -64,7 +64,7 @@ or within Neuroglancer, press `h` or click on the button labeled `?` in the uppe
 
 - Hover over a segmentation layer name to see the current list of objects shown and to access the opacity sliders.
 
-- Hover over an image layer name to access the opacity slider and the text editor for modifying the [rendering code](src/neuroglancer/sliceview/image_layer_rendering.md).
+- Hover over an image layer name to access the opacity slider and the text editor for modifying the [rendering code](src/sliceview/image_layer_rendering.md).
 
 # Troubleshooting
 
@@ -111,11 +111,11 @@ The "frontend" UI thread handles user actions and rendering, while the "backend"
 
 # Documentation Index
 
-- [Image Layer Rendering](src/neuroglancer/sliceview/image_layer_rendering.md)
-- [Cross-sectional view implementation architecture](src/neuroglancer/sliceview/README.md)
-- [Compressed segmentation format](src/neuroglancer/sliceview/compressed_segmentation/README.md)
-- [Data chunk management](src/neuroglancer/chunk_manager/)
-- [On-GPU hashing](src/neuroglancer/gpu_hash/)
+- [Image Layer Rendering](src/sliceview/image_layer_rendering.md)
+- [Cross-sectional view implementation architecture](src/sliceview/README.md)
+- [Compressed segmentation format](src/sliceview/compressed_segmentation/README.md)
+- [Data chunk management](src/chunk_manager/)
+- [On-GPU hashing](src/gpu_hash/)
 
 # Building
 
@@ -154,7 +154,7 @@ https://github.com/creationix/nvm
 
    For example,
 
-   `npm test -- --pattern='src/neuroglancer/util/uint64*'`
+   `npm test -- --pattern='src/util/uint64*'`
 
 6. See [package.json](package.json) for other commands available.
 
@@ -171,11 +171,11 @@ There is a Google Group/mailing list for discussion related to Neuroglancer:
   Docker deployment of the [Neuroglancer Python integration](python/README.md).
 - [FZJ-INM1-BDA/neuroglancer-scripts](https://github.com/FZJ-INM1-BDA/neuroglancer-scripts) -
   Scripts for converting the [BigBrain](https://bigbrain.loris.ca) dataset to the
-  Neuroglancer [precomputed data format](src/neuroglancer/datasource/precomputed), which may serve
+  Neuroglancer [precomputed data format](src/datasource/precomputed), which may serve
   as a useful example for converting other datasets.
 - [BigArrays.jl](https://github.com/seung-lab/BigArrays.jl) - Julia interface of neuroglancer precomputed data format.
 - [cloudvolume](https://github.com/seung-lab/cloud-volume) - Python interface of neuroglancer precomputed data format.
-- [multiresolution-mesh-creator](https://github.com/janelia-cosem/multiresolution-mesh-creator) - Python tool for creating [multi-resolution meshes](https://github.com/google/neuroglancer/blob/master/src/neuroglancer/datasource/precomputed/meshes.md#multi-resolution-mesh-format) from single resolution - or multiscale - meshes.
+- [multiresolution-mesh-creator](https://github.com/janelia-cosem/multiresolution-mesh-creator) - Python tool for creating [multi-resolution meshes](https://github.com/google/neuroglancer/blob/master/src/datasource/precomputed/meshes.md#multi-resolution-mesh-format) from single resolution - or multiscale - meshes.
 - [Igneous](https://github.com/seung-lab/igneous) - Python pipeline for scalable meshing, skeletonizing, downsampling, and managment of large 3d images focusing on Neuroglancer Precomputed format.
 
 # Contributing
diff --git a/python/ext/src/compress_segmentation.h b/python/ext/src/compress_segmentation.h
index e436060c1..67453296d 100644
--- a/python/ext/src/compress_segmentation.h
+++ b/python/ext/src/compress_segmentation.h
@@ -15,7 +15,7 @@
  */
 
 // Implements encoding into the compressed segmentation format described at
-// https://github.com/google/neuroglancer/tree/master/src/neuroglancer/sliceview/compressed_segmentation.
+// https://github.com/google/neuroglancer/tree/master/src/sliceview/compressed_segmentation.
 //
 // Only uint32 and uint64 volumes are supported.
 
diff --git a/python/neuroglancer/segment_colors.py b/python/neuroglancer/segment_colors.py
index 2751b77c1..3c51e6595 100644
--- a/python/neuroglancer/segment_colors.py
+++ b/python/neuroglancer/segment_colors.py
@@ -17,7 +17,7 @@
 
 def hash_function(state, value):
     """Python implementation of hashCombine() function
-    in src/neuroglancer/gpu_hash/hash_function.ts,
+    in src/gpu_hash/hash_function.ts,
     a modified murmur hash
     """
     k1 = 0xCC9E2D51
@@ -34,7 +34,7 @@ def hash_function(state, value):
 
 def hsv_to_rgb(h, s, v):
     """Convert H,S,V values to RGB values.
-    Python implementation of hsvToRgb in src/neuroglancer/util/colorspace.ts"""
+    Python implementation of hsvToRgb in src/util/colorspace.ts"""
     h *= 6
     hue_index = math.floor(h)
     remainder = h - hue_index
@@ -59,7 +59,7 @@ def hsv_to_rgb(h, s, v):
 def pack_color(rgb_vec):
     """Returns an integer formed
     by concatenating the channels of the input color vector.
-    Python implementation of packColor in src/neuroglancer/util/color.ts
+    Python implementation of packColor in src/util/color.ts
     """
     result = 0
     for i in range(len(rgb_vec)):
diff --git a/src/annotation/user_layer.ts b/src/annotation/user_layer.ts
index e67b863b4..edc9c1fc5 100644
--- a/src/annotation/user_layer.ts
+++ b/src/annotation/user_layer.ts
@@ -799,7 +799,7 @@ class RenderingOptionsTab extends Tab {
     topRow.appendChild(
       makeHelpButton({
         title: "Documentation on annotation rendering",
-        href: "https://github.com/google/neuroglancer/blob/master/src/neuroglancer/annotation/rendering.md",
+        href: "https://github.com/google/neuroglancer/blob/master/src/annotation/rendering.md",
       }),
     );
     element.appendChild(topRow);
diff --git a/src/datasource/precomputed/volume.md b/src/datasource/precomputed/volume.md
index f17715fbd..46b1a54c8 100644
--- a/src/datasource/precomputed/volume.md
+++ b/src/datasource/precomputed/volume.md
@@ -24,7 +24,7 @@ The root value must be a JSON object with the following members:
   permits automatically detecting paths to volumes, meshes, and skeletons.
 - `"type"`: One of `"image"` or `"segmentation"`, specifying the type of the volume.
 - `"data_type"`: A string value equal (case-insensitively) to the name of one of the supported
-  `DataType` values specified in [data_type.ts](/src/neuroglancer/util/data_type.ts). May be one of
+  `DataType` values specified in [data_type.ts](/src/util/data_type.ts). May be one of
   `"uint8"`, `"uint16"`, `"uint32"`, `"uint64"`, or `"float32"`. `"float32"` should only be specified
   for `"image"` volumes.
 - `"num_channels"`: An integer value specifying the number of channels in the volume. Must be `1`
@@ -192,7 +192,7 @@ of the image corresponds to the flattened `[x, y, z]` Fortran-order representati
 ### compressed_segmentation chunk encoding
 
 The subvolume data for the chunk is encoded using the multi-channel
-format [compressed segmentation format](/src/neuroglancer/sliceview/compressed_segmentation). The
+format [compressed segmentation format](/src/sliceview/compressed_segmentation). The
 `"data_type"` must be either `"uint32"` or `"uint64"`. The compression block size is specified by
 the `"compressed_segmentation_block_size"` property in the `info` JSON file.
 
diff --git a/src/gpu_hash/README.md b/src/gpu_hash/README.md
index be2d1d3d4..96a2b7978 100644
--- a/src/gpu_hash/README.md
+++ b/src/gpu_hash/README.md
@@ -1,6 +1,6 @@
 This directory implements hash functions, hash sets, and hash maps that support efficient use from
 both JavaScript and from WebGL shaders running on the GPU. These are used to
-perform [pseudo-random color mapping of uint64 object IDs](/src/neuroglancer/segment_color.ts),
+perform [pseudo-random color mapping of uint64 object IDs](/src/segment_color.ts),
 highlighting of a set of object ids, and equivalence maps over object ids.
 
 The design is heavily constrained by the limitations of WebGL 1.0 (OpenGL ES Shading Language 1.0) shaders:
diff --git a/src/image_user_layer.ts b/src/image_user_layer.ts
index 8b64fb20f..8e1544a0d 100644
--- a/src/image_user_layer.ts
+++ b/src/image_user_layer.ts
@@ -500,7 +500,7 @@ class RenderingOptionsTab extends Tab {
     topRow.appendChild(
       makeHelpButton({
         title: "Documentation on image layer rendering",
-        href: "https://github.com/google/neuroglancer/blob/master/src/neuroglancer/sliceview/image_layer_rendering.md",
+        href: "https://github.com/google/neuroglancer/blob/master/src/sliceview/image_layer_rendering.md",
       }),
     );
 
diff --git a/src/mesh/frontend.ts b/src/mesh/frontend.ts
index 9bce833db..18ff98189 100644
--- a/src/mesh/frontend.ts
+++ b/src/mesh/frontend.ts
@@ -79,7 +79,7 @@ const tempMat3 = mat3.create();
 
 // To validate the octrees and to determine the multiscale fragment responsible for each framebuffer
 // location, set `DEBUG_MULTISCALE_FRAGMENTS=true` and also set `DEBUG_PICKING=true` in
-// `src/neuroglancer/object_picking.ts`.
+// `src/object_picking.ts`.
 const DEBUG_MULTISCALE_FRAGMENTS = false;
 
 function copyMeshDataToGpu(
diff --git a/src/single_mesh_user_layer.ts b/src/single_mesh_user_layer.ts
index 48c8ee937..ef50916c0 100644
--- a/src/single_mesh_user_layer.ts
+++ b/src/single_mesh_user_layer.ts
@@ -216,7 +216,7 @@ class DisplayOptionsTab extends Tab {
     topRow.appendChild(
       makeHelpButton({
         title: "Documentation on single mesh layer rendering",
-        href: "https://github.com/google/neuroglancer/blob/master/src/neuroglancer/sliceview/image_layer_rendering.md",
+        href: "https://github.com/google/neuroglancer/blob/master/src/sliceview/image_layer_rendering.md",
       }),
     );
 
diff --git a/src/ui/segmentation_display_options_tab.ts b/src/ui/segmentation_display_options_tab.ts
index 855fdd5bf..62ec572cb 100644
--- a/src/ui/segmentation_display_options_tab.ts
+++ b/src/ui/segmentation_display_options_tab.ts
@@ -93,7 +93,7 @@ export class DisplayOptionsTab extends Tab {
           topRow.appendChild(
             makeHelpButton({
               title: "Documentation on skeleton rendering",
-              href: "https://github.com/google/neuroglancer/blob/master/src/neuroglancer/sliceview/image_layer_rendering.md",
+              href: "https://github.com/google/neuroglancer/blob/master/src/sliceview/image_layer_rendering.md",
             }),
           );
           parent.appendChild(topRow);
diff --git a/testdata/generate_npy_examples.py b/testdata/generate_npy_examples.py
index 265f4afe4..387395d8c 100755
--- a/testdata/generate_npy_examples.py
+++ b/testdata/generate_npy_examples.py
@@ -15,7 +15,7 @@
 # limitations under the License.
 
 # This file generates npy_test.*.{npy,json} files for use by
-# src/neuroglancer/util/npy.spec.ts.
+# src/util/npy.spec.ts.
 #
 # This should be run from within the testdata/ directory.