diff --git a/docs/computation/execute.md b/docs/computation/execute.md
index 0b280224..a2cde2a6 100644
--- a/docs/computation/execute.md
+++ b/docs/computation/execute.md
@@ -23,7 +23,8 @@ To trigger the execution of notebook pages, use the global `nb_execution_mode` c
 | -------- | -------------------------------------------------------------------- |
 | `off`    | Do not execute the notebook                                          |
 | `force`  | Always execute the notebook (before parsing)                         |
-| `auto`   | Execute notebooks with missing outputs (before parsing)              |
+| `auto`   | Execute notebooks with any missing outputs (before parsing)          |
+| `lazy`   | Execute notebooks without any existing outputs (before parsing)      |
 | `cache`  | Execute notebook and store/retrieve outputs from a cache             |
 | `inline` | Execute the notebook during parsing (allows for variable evaluation) |
 
@@ -36,6 +37,12 @@ nb_execution_mode = "auto"
 This will only execute notebooks that are missing at least one output.
 If a notebook has *all* of its outputs populated, then it will not be executed.
 
+To only execute notebooks that do not have *any* of its outputs populated, change the above configuration value to:
+
+```python
+nb_execution_mode = "lazy"
+```
+
 To force the execution of all notebooks, regardless of their outputs, change the above configuration value to:
 
 ```python
diff --git a/docs/conf.py b/docs/conf.py
index 0f954923..6fa1b3d2 100644
--- a/docs/conf.py
+++ b/docs/conf.py
@@ -95,6 +95,7 @@
         "off",
         "force",
         "auto",
+        "lazy",
         "cache",
         "commonmark",
         "gfm",
diff --git a/myst_nb/core/config.py b/myst_nb/core/config.py
index 023cccac..6c2ff456 100644
--- a/myst_nb/core/config.py
+++ b/myst_nb/core/config.py
@@ -197,13 +197,16 @@ def __post_init__(self):
             "sections": (Section.global_lvl, Section.file_lvl, Section.execute),
         },
     )
-    execution_mode: Literal["off", "force", "auto", "cache", "inline"] = dc.field(
+    execution_mode: Literal[
+        "off", "force", "auto", "lazy", "cache", "inline"
+    ] = dc.field(
         default="auto",
         metadata={
             "validator": in_(
                 [
                     "off",
                     "auto",
+                    "lazy",
                     "force",
                     "cache",
                     "inline",
diff --git a/myst_nb/core/execute/__init__.py b/myst_nb/core/execute/__init__.py
index 5945988a..dc559982 100644
--- a/myst_nb/core/execute/__init__.py
+++ b/myst_nb/core/execute/__init__.py
@@ -50,7 +50,8 @@ def create_client(
                 logger.info(f"Excluded from execution by pattern: {pattern!r}")
                 return NotebookClientBase(notebook, path, nb_config, logger)
 
-    # 'auto' mode only executes the notebook if it is missing at least one output
+    # 'auto' mode only executes the notebook if it is missing at least one output,
+    # 'lazy' mode only executes if all outputs are missing.
     missing_outputs = (
         len(cell.outputs) == 0 for cell in notebook.cells if cell["cell_type"] == "code"
     )
@@ -58,7 +59,11 @@ def create_client(
         logger.info("Skipped execution in 'auto' mode (all outputs present)")
         return NotebookClientBase(notebook, path, nb_config, logger)
 
-    if nb_config.execution_mode in ("auto", "force"):
+    if nb_config.execution_mode == "lazy" and not all(missing_outputs):
+        logger.info("Skipped execution in 'lazy' mode (at least one output present)")
+        return NotebookClientBase(notebook, path, nb_config, logger)
+
+    if nb_config.execution_mode in ("auto", "lazy", "force"):
         return NotebookClientDirect(notebook, path, nb_config, logger)
 
     if nb_config.execution_mode == "cache":
diff --git a/tests/test_execute.py b/tests/test_execute.py
index 9597182a..f763135a 100644
--- a/tests/test_execute.py
+++ b/tests/test_execute.py
@@ -52,6 +52,21 @@ def test_basic_unrun_auto(sphinx_run, file_regression, check_nbs):
     assert data["succeeded"] is True
 
 
+@pytest.mark.sphinx_params("basic_unrun.ipynb", conf={"nb_execution_mode": "lazy"})
+def test_basic_unrun_lazy(sphinx_run, file_regression, check_nbs):
+    sphinx_run.build()
+    # print(sphinx_run.status())
+    assert sphinx_run.warnings() == ""
+    assert "test_name" in sphinx_run.app.env.metadata["basic_unrun"]
+    regress_nb_doc(file_regression, sphinx_run, check_nbs)
+
+    assert NbMetadataCollector.new_exec_data(sphinx_run.env)
+    data = NbMetadataCollector.get_exec_data(sphinx_run.env, "basic_unrun")
+    assert data
+    assert data["method"] == "lazy"
+    assert data["succeeded"] is True
+
+
 @pytest.mark.sphinx_params("basic_unrun.ipynb", conf={"nb_execution_mode": "cache"})
 def test_basic_unrun_cache(sphinx_run, file_regression, check_nbs):
     """The outputs should be populated."""
@@ -149,6 +164,21 @@ def test_basic_failing_auto(sphinx_run, file_regression, check_nbs):
     sphinx_run.get_report_file()
 
 
+@pytest.mark.skipif(ipy_version[0] < 8, reason="Error message changes for ipython v8")
+@pytest.mark.sphinx_params("basic_failing.ipynb", conf={"nb_execution_mode": "lazy"})
+def test_basic_failing_lazy(sphinx_run, file_regression, check_nbs):
+    sphinx_run.build()
+    assert "Executing notebook failed" in sphinx_run.warnings()
+    regress_nb_doc(file_regression, sphinx_run, check_nbs)
+
+    data = NbMetadataCollector.get_exec_data(sphinx_run.env, "basic_failing")
+    assert data
+    assert data["method"] == "lazy"
+    assert data["succeeded"] is False
+    assert data["traceback"]
+    sphinx_run.get_report_file()
+
+
 @pytest.mark.skipif(ipy_version[0] < 8, reason="Error message changes for ipython v8")
 @pytest.mark.sphinx_params("basic_failing.ipynb", conf={"nb_execution_mode": "inline"})
 def test_basic_failing_inline(sphinx_run, file_regression, check_nbs):