diff --git a/docs/cli/pipeline-approve.rst b/docs/cli/pipeline-approve.rst
index 922b17d..45f3467 100644
--- a/docs/cli/pipeline-approve.rst
+++ b/docs/cli/pipeline-approve.rst
@@ -16,7 +16,10 @@ Usage
    toasty pipeline approve [--workdir=WORKDIR] {IMAGE-IDs...}
 
 The ``IMAGE-IDs`` argument specifies one or more images by their unique
-identifiers.
+identifiers. You can specify exact ID’s, or `glob patterns`_ as processed by the
+Python ``fnmatch`` module. See examples below.
+
+.. _glob patterns: https://docs.python.org/3/library/fnmatch.html#module-fnmatch
 
 The ``WORKDIR`` argument optionally specifies the location of the pipeline
 workspace directory. The default is the current directory.
@@ -51,6 +54,18 @@ If everything is OK, you can mark the image as approved:
 
    toasty pipeline approve noao0201b
 
+You can use `glob patterns`_ to match image names. For instance,
+
+.. code-block:: shell
+
+   toasty pipeline approve "vla*20" "?vlba"
+
+will match every processed image whose identifier begins with ``vla`` and ends
+with ``20``, as well as those whose names are exactly four letters long and end
+with ``vlba``. You generally must make sure to encase glob arguments in
+quotation marks, as shown above, to prevent your shell from attempting to
+process them before Toasty gets a chance to.
+
 After approval of a batch of images, the next step is to :ref:`cli-pipeline-publish`.
 
 
diff --git a/docs/cli/pipeline-fetch.rst b/docs/cli/pipeline-fetch.rst
index 718c22c..7d8af1a 100644
--- a/docs/cli/pipeline-fetch.rst
+++ b/docs/cli/pipeline-fetch.rst
@@ -16,7 +16,10 @@ Usage
    toasty pipeline fetch [--workdir=WORKDIR] {IMAGE-IDs...}
 
 The ``IMAGE-IDs`` argument specifies one or more images by their unique
-identifiers.
+identifiers. You can specify exact ID’s, or `glob patterns`_ as processed by the
+Python ``fnmatch`` module. See examples below.
+
+.. _glob patterns: https://docs.python.org/3/library/fnmatch.html#module-fnmatch
 
 The ``WORKDIR`` argument optionally specifies the location of the pipeline
 workspace directory. The default is the current directory.
@@ -34,6 +37,22 @@ Fetch two images:
 After fetching, the next step is to :ref:`cli-pipeline-process-todos`.
 
 
+Example
+=======
+
+You can use `glob patterns`_ to match candidate names. For instance,
+
+.. code-block:: shell
+
+   toasty pipeline fetch "rubin-*" "soar?"
+
+will match every candidate whose name begins with ``rubin-``,  as well as those
+whose names are exactly five letters long and start with ``soar``. You generally
+must make sure to encase glob arguments in quotation marks, as shown above, to
+prevent your shell from attempting to process them before Toasty gets a chance
+to.
+
+
 Notes
 =====
 
diff --git a/toasty/pipeline/cli.py b/toasty/pipeline/cli.py
index 2f08bdb..ce6a497 100644
--- a/toasty/pipeline/cli.py
+++ b/toasty/pipeline/cli.py
@@ -12,6 +12,8 @@ pipeline_impl
 '''.split()
 
 import argparse
+from fnmatch import fnmatch
+import glob
 import os.path
 import sys
 
@@ -19,6 +21,34 @@ from ..cli import die, warn
 from . import NotActionableError
 
 
+def evaluate_imageid_args(searchdir, args):
+    """
+    Figure out which image-ID's to process.
+    """
+
+    matched_ids = set()
+    globs_todo = set()
+
+    for arg in args:
+        if glob.has_magic(arg):
+            globs_todo.add(arg)
+        else:
+            # If an ID is explicitly (non-gobbily) added, always add it to the
+            # list, without checking if it exists in `searchdir`. We could check
+            # for it in searchdir now, but we'll have to check later anyway, so
+            # we don't bother.
+            matched_ids.add(arg)
+
+    if len(globs_todo):
+        for filename in os.listdir(searchdir):
+            for g in globs_todo:
+                if fnmatch(filename, g):
+                    matched_ids.add(filename)
+                    break
+
+    return sorted(matched_ids)
+
+
 # The "approve" subcommand
 
 def approve_setup_parser(parser):
@@ -31,8 +61,8 @@ def approve_setup_parser(parser):
     parser.add_argument(
         'cand_ids',
         nargs = '+',
-        metavar = 'CAND-ID',
-        help = 'Name(s) of candidate(s) to approve and prepare for processing'
+        metavar = 'IMAGE-ID',
+        help = 'Name(s) of image(s) to approve for publication (globs accepted)'
     )
 
 
@@ -51,7 +81,7 @@ def approve_impl(settings):
     proc_dir = mgr._ensure_dir('processed')
     app_dir = mgr._ensure_dir('approved')
 
-    for cid in settings.cand_ids:
+    for cid in evaluate_imageid_args(proc_dir, settings.cand_ids):
         if not os.path.isdir(os.path.join(proc_dir, cid)):
             die(f'no such processed candidate ID {cid!r}')
 
@@ -90,7 +120,7 @@ def fetch_setup_parser(parser):
         'cand_ids',
         nargs = '+',
         metavar = 'CAND-ID',
-        help = 'Name(s) of candidate(s) to fetch and prepare for processing'
+        help = 'Name(s) of candidate(s) to fetch and prepare for processing (globs accepted)'
     )
 
 
@@ -102,7 +132,7 @@ def fetch_impl(settings):
     rej_dir = mgr._ensure_dir('rejects')
     src = mgr.get_image_source()
 
-    for cid in settings.cand_ids:
+    for cid in evaluate_imageid_args(cand_dir, settings.cand_ids):
         # Funky structure here is to try to ensure that cdata is closed in case
         # a NotActionable happens, so that we can move the directory on Windows.
         try: