diff --git a/docs/guides/README.md b/docs/guides/README.md
index 4753f7e22..7101a5d81 100644
--- a/docs/guides/README.md
+++ b/docs/guides/README.md
@@ -2,7 +2,8 @@
 
 This folder contains various task-specific curation guides.
 
-- [Curating new providers](/curation/providers)
+- [Curating new providers](curation/providers)
+- [Curating new publications and references](curation/publications)
 
 ## How to add new guides
 
@@ -11,7 +12,7 @@ This folder contains various task-specific curation guides.
    (see
    [here](https://github.com/biopragmatics/bioregistry/blob/fe2a685503ae2c9ff863908bf885c71fd240c21d/docs/guides/providers.md?plain=1#L1-L5)
    for an example)
-3. Add it to the list above
+3. Add it to the list above. Don't include a forward slash `/` in the beginning of the link!
 
 ## What makes a good guide
 
diff --git a/docs/guides/publications.md b/docs/guides/publications.md
new file mode 100644
index 000000000..6f3001b38
--- /dev/null
+++ b/docs/guides/publications.md
@@ -0,0 +1,76 @@
+---
+layout: page
+title: Curating Publications and References
+permalink: /curation/publications
+---
+
+The example below shows a subset of the record for
+[3D Metabolites (3dmet)](https://bioregistry.io/3dmet) that highlights the `publications` list.
+Note that each entry is a dictionary with several parts:
+
+1. `title` (required) - the title of the paper
+2. `year` (highly recommended) - the year of publication of the paper
+3. `pubmed`, `doi`, and `pmc` (one or more required) - identifiers for the paper
+
+```json
+"3dmet": {
+    "name": "3D Metabolites",
+    "publications": [
+      {
+        "doi": "10.2142/biophysico.15.0_87",
+        "pmc": "PMC5992871",
+        "pubmed": "29892514",
+        "title": "Chemical curation to improve data accuracy: recent development of the 3DMET database",
+        "year": 2018
+      },
+      {
+        "doi": "10.1021/ci300309k",
+        "pubmed": "23293959",
+        "title": "Three-dimensional structure database of natural metabolites (3DMET): a novel database of curated 3D structures",
+        "year": 2013
+      }
+    ]
+  },
+```
+
+Similarly, there are URL references that are not _publications_ that are worth curating. These can be
+stored in the `references` list. For example, the 
+[Registry of Toxic Effects of Chemical Substances (rtecs)](https://bioregistry.io/rtecs) entry appears in the
+Bioregistry because of its usage, but it is hard to find information on the internet about it. Therefore, the
+references list is perfect for storing references to PDFs and webpages that describe the resource.
+
+```json
+"rtecs": {
+   "name": "Registry of Toxic Effects of Chemical Substances",
+   "publications": [
+   {
+      "doi": "10.1016/s1074-9098%2899%2900058-1",
+      "title": "An overview of the Registry of Toxic Effects of Chemical Substances (RTECS): Critical information on chemical hazards",
+      "year": 1999
+   }
+   ],
+   "references": [
+      "https://www.cdc.gov/niosh/docs/97-119/pdfs/97-119.pdf",
+      "https://www.cdc.gov/niosh/npg/npgdrtec.html"
+   ]
+}
+```
+
+What else is good to keep track of in the references list:
+
+1. Bioregistry issue or pull requests about the resource
+2. Links to webpages describing the identifier resource
+3. Links to discussions on Slack or other platforms (keeping in mind links might not last forever)
+4. Any other context that's useful for a Bioregistry reader
+
+## Why Should I Curate Publications and References?
+
+1. They give additional context for Bioregistry readers who want to know more about the paper
+2. They make it easier to attribute usage of identifiers from a given resource to its authors
+3. They enable global landscape analysis of when and where identifier resources are being made. The following image is
+   automatically regenerated with each Bioregistry update:
+
+   ![](https://raw.githubusercontent.com/biopragmatics/bioregistry/refs/heads/main/docs/img/bibliography_years.svg)
+4. They support the training of a machine learning for semi-automated curation of additional literature. See
+   this [talk](https://docs.google.com/presentation/d/1h2IajyGkUxUPHubEi8_WE6xW6TOuOihn5zsmi4kYrrc/edit?usp=sharing)
+   from the 2022 Workshop on Prefixes, CURIEs, and IRIs.
diff --git a/tests/test_data.py b/tests/test_data.py
index ebee8d48c..3496a674a 100644
--- a/tests/test_data.py
+++ b/tests/test_data.py
@@ -854,13 +854,18 @@ def test_request_issue(self):
 
     def test_publications(self):
         """Test references and publications are sorted right."""
+        msg_fmt = (
+            "Rather than writing a {} link in the `references` list, "
+            "you should encode it in the `publications` instead. "
+            "See https://biopragmatics.github.io/bioregistry/curation/publications for help."
+        )
         for prefix, resource in self.registry.items():
             with self.subTest(prefix=prefix):
                 if resource.references:
                     for reference in resource.references:
-                        self.assertNotIn("doi", reference)
-                        self.assertNotIn("pubmed", reference)
-                        self.assertNotIn("pmc", reference)
+                        self.assertNotIn("doi", reference, msg=msg_fmt.format("DOI"))
+                        self.assertNotIn("pubmed", reference, msg=msg_fmt.format("PubMed"))
+                        self.assertNotIn("pmc", reference, msg_fmt.format("PMC"))
                         self.assertNotIn("arxiv", reference)
                 if resource.publications:
                     for publication in resource.publications: