diff --git a/.gitignore b/.gitignore
index 90f4a10ed5f..686c7efa701 100644
--- a/.gitignore
+++ b/.gitignore
@@ -5,8 +5,9 @@ __pycache__
 .hypothesis/
 
 # temp files from docs build
+doc/*.nc
 doc/auto_gallery
-doc/example.nc
+doc/rasm.zarr
 doc/savefig
 
 # C extensions
@@ -72,4 +73,3 @@ xarray/tests/data/*.grib.*.idx
 Icon*
 
 .ipynb_checkpoints
-doc/rasm.zarr
diff --git a/doc/conf.py b/doc/conf.py
index 93174c6aaec..bc56033891a 100644
--- a/doc/conf.py
+++ b/doc/conf.py
@@ -28,9 +28,9 @@
 print("python exec:", sys.executable)
 print("sys.path:", sys.path)
 
-if "conda" in sys.executable:
+if "CONDA_DEFAULT_ENV" in os.environ or "conda" in sys.executable:
     print("conda environment:")
-    subprocess.run(["conda", "list"])
+    subprocess.run([os.environ.get("CONDA_EXE", "conda"), "list"])
 else:
     print("pip environment:")
     subprocess.run([sys.executable, "-m", "pip", "list"])
diff --git a/doc/getting-started-guide/quick-overview.rst b/doc/getting-started-guide/quick-overview.rst
index 5bb5bb88ad3..bdd4e0df413 100644
--- a/doc/getting-started-guide/quick-overview.rst
+++ b/doc/getting-started-guide/quick-overview.rst
@@ -215,13 +215,15 @@ You can directly read and write xarray objects to disk using :py:meth:`~xarray.D
 .. ipython:: python
 
     ds.to_netcdf("example.nc")
-    xr.open_dataset("example.nc")
+    reopened = xr.open_dataset("example.nc")
+    reopened
 
 .. ipython:: python
     :suppress:
 
     import os
 
+    reopened.close()
     os.remove("example.nc")
 
 
diff --git a/doc/internals/zarr-encoding-spec.rst b/doc/internals/zarr-encoding-spec.rst
index 082d7984f59..6f8501a6b11 100644
--- a/doc/internals/zarr-encoding-spec.rst
+++ b/doc/internals/zarr-encoding-spec.rst
@@ -63,3 +63,10 @@ re-open it directly with Zarr:
     print(os.listdir("rasm.zarr"))
     print(zgroup.tree())
     dict(zgroup["Tair"].attrs)
+
+.. ipython:: python
+    :suppress:
+
+    import shutil
+
+    shutil.rmtree("rasm.zarr")
diff --git a/doc/user-guide/dask.rst b/doc/user-guide/dask.rst
index 4998cc68828..ebbbaa199f6 100644
--- a/doc/user-guide/dask.rst
+++ b/doc/user-guide/dask.rst
@@ -55,6 +55,8 @@ argument to :py:func:`~xarray.open_dataset` or using the
 .. ipython:: python
     :suppress:
 
+    import os
+
     import numpy as np
     import pandas as pd
     import xarray as xr
@@ -129,6 +131,11 @@ will return a ``dask.delayed`` object that can be computed later.
     with ProgressBar():
         results = delayed_obj.compute()
 
+.. ipython:: python
+    :suppress:
+
+    os.remove("manipulated-example-data.nc")  # Was not opened.
+
 .. note::
 
     When using Dask's distributed scheduler to write NETCDF4 files,
@@ -147,13 +154,6 @@ A dataset can also be converted to a Dask DataFrame using :py:meth:`~xarray.Data
 
 Dask DataFrames do not support multi-indexes so the coordinate variables from the dataset are included as columns in the Dask DataFrame.
 
-.. ipython:: python
-    :suppress:
-
-    import os
-
-    os.remove("example-data.nc")
-    os.remove("manipulated-example-data.nc")
 
 Using Dask with xarray
 ----------------------
@@ -210,7 +210,7 @@ Dask arrays using the :py:meth:`~xarray.Dataset.persist` method:
 
 .. ipython:: python
 
-    ds = ds.persist()
+    persisted = ds.persist()
 
 :py:meth:`~xarray.Dataset.persist` is particularly useful when using a
 distributed cluster because the data will be loaded into distributed memory
@@ -232,11 +232,6 @@ chunk size depends both on your data and on the operations you want to perform.
 With xarray, both converting data to a Dask arrays and converting the chunk
 sizes of Dask arrays is done with the :py:meth:`~xarray.Dataset.chunk` method:
 
-.. ipython:: python
-    :suppress:
-
-    ds = ds.chunk({"time": 10})
-
 .. ipython:: python
 
     rechunked = ds.chunk({"latitude": 100, "longitude": 100})
@@ -508,6 +503,11 @@ Notice that the 0-shaped sizes were not printed to screen. Since ``template`` ha
     expected = ds + 10 + 10
     mapped.identical(expected)
 
+.. ipython:: python
+    :suppress:
+
+    ds.close()  # Closes "example-data.nc".
+    os.remove("example-data.nc")
 
 .. tip::
 
diff --git a/doc/user-guide/io.rst b/doc/user-guide/io.rst
index 16b8708231e..713b5d7382b 100644
--- a/doc/user-guide/io.rst
+++ b/doc/user-guide/io.rst
@@ -11,6 +11,8 @@ format (recommended).
 .. ipython:: python
     :suppress:
 
+    import os
+
     import numpy as np
     import pandas as pd
     import xarray as xr
@@ -84,6 +86,13 @@ We can load netCDF files to create a new Dataset using
     ds_disk = xr.open_dataset("saved_on_disk.nc")
     ds_disk
 
+.. ipython:: python
+    :suppress:
+
+    # Close "saved_on_disk.nc", but retain the file until after closing or deleting other
+    # datasets that will refer to it.
+    ds_disk.close()
+
 Similarly, a DataArray can be saved to disk using the
 :py:meth:`DataArray.to_netcdf` method, and loaded
 from disk using the :py:func:`open_dataarray` function. As netCDF files
@@ -204,11 +213,6 @@ You can view this encoding information (among others) in the
 Note that all operations that manipulate variables other than indexing
 will remove encoding information.
 
-.. ipython:: python
-    :suppress:
-
-    ds_disk.close()
-
 
 .. _combining multiple files:
 
@@ -484,13 +488,13 @@ and currently raises a warning unless ``invalid_netcdf=True`` is set:
     da.to_netcdf("complex.nc", engine="h5netcdf", invalid_netcdf=True)
 
     # Reading it back
-    xr.open_dataarray("complex.nc", engine="h5netcdf")
+    reopened = xr.open_dataarray("complex.nc", engine="h5netcdf")
+    reopened
 
 .. ipython:: python
     :suppress:
 
-    import os
-
+    reopened.close()
     os.remove("complex.nc")
 
 .. warning::
@@ -723,16 +727,19 @@ To export just the dataset schema without the data itself, use the
 
     ds.to_dict(data=False)
 
-This can be useful for generating indices of dataset contents to expose to
-search indices or other automated data discovery tools.
-
 .. ipython:: python
     :suppress:
 
-    import os
-
+    # We're now done with the dataset named `ds`.  Although the `with` statement closed
+    # the dataset, displaying the unpickled pickle of `ds` re-opened "saved_on_disk.nc".
+    # However, `ds` (rather than the unpickled dataset) refers to the open file.  Delete
+    # `ds` to close the file.
+    del ds
     os.remove("saved_on_disk.nc")
 
+This can be useful for generating indices of dataset contents to expose to
+search indices or other automated data discovery tools.
+
 .. _io.rasterio:
 
 Rasterio
diff --git a/doc/user-guide/weather-climate.rst b/doc/user-guide/weather-climate.rst
index 893e7b50429..6db0396ee4e 100644
--- a/doc/user-guide/weather-climate.rst
+++ b/doc/user-guide/weather-climate.rst
@@ -218,13 +218,15 @@ For data indexed by a :py:class:`~xarray.CFTimeIndex` xarray currently supports:
 .. ipython:: python
 
     da.to_netcdf("example-no-leap.nc")
-    xr.open_dataset("example-no-leap.nc")
+    reopened = xr.open_dataset("example-no-leap.nc")
+    reopened
 
 .. ipython:: python
     :suppress:
 
     import os
 
+    reopened.close()
     os.remove("example-no-leap.nc")
 
 - And resampling along the time dimension for data indexed by a :py:class:`~xarray.CFTimeIndex`:
diff --git a/doc/whats-new.rst b/doc/whats-new.rst
index 93271cca3e4..f7249f192bd 100644
--- a/doc/whats-new.rst
+++ b/doc/whats-new.rst
@@ -38,6 +38,9 @@ Bug fixes
 Documentation
 ~~~~~~~~~~~~~
 
+- Delete files of datasets saved to disk while building the documentation and enable
+  building on Windows via `sphinx-build` (:pull:`6237`).
+  By `Stan West <https://github.com/stanwest>`_.
 
 Internal Changes
 ~~~~~~~~~~~~~~~~