diff --git a/event_model/scrub.py b/event_model/scrub.py
new file mode 100644
index 000000000..ff6bb1107
--- /dev/null
+++ b/event_model/scrub.py
@@ -0,0 +1,99 @@
+import event_model
+import itertools
+
+
+def scrub(hdrs, new_epoch, start_scan_id, filter_start=lambda x: x):
+    """
+    Scrubs a set of headers of most identifying information.
+
+    Leaves the internal time and scan_id differences between the headers intact.
+
+    Parameters
+    ----------
+    hdrs : Iterable[BlueskyRun]
+        The headers to sanitize
+
+    new_epoch : float
+        The first header in hdrs is deemed to be at this time, all other headers
+        will have their times adjusted so that the relative time is the same.
+
+    start_scan_id : int
+
+       The first header is deemed to be at this scan_id, the rest of the
+       scan_ids will be adjusted accordingly.
+
+    Yields
+    ------
+    name : str
+    doc : dict
+
+    """
+    time_offset = None
+    scan_id_offset = None
+
+    for h, first in zip(hdrs, itertools.chain([True], itertools.repeat(False))):
+        if first:
+            time_offset = h.metadata["start"]["time"] - new_epoch
+            scan_id_offset = h.metadata["start"]["scan_id"] - start_scan_id
+
+        docs = h.documents(fill="no")
+
+        _, start = next(docs)
+        start = dict(start)
+        start = filter_start(start)
+        start.pop("uid")
+
+        new_time = start.pop("time") - time_offset
+        new_scan_id = start.pop("scan_id", 0) - scan_id_offset
+
+        desc_map = {}
+        run_bundle = event_model.compose_run(
+            time=new_time, metadata={**start, "scan_id": new_scan_id}
+        )
+        yield "start", run_bundle.start_doc
+
+        for name, doc in docs:
+            if name == "datum_page":
+                yield name, doc
+            elif name == "resource":
+                res = dict(doc)
+                res["run_start"] = run_bundle.start_doc["uid"]
+                yield name, res
+            elif name == "descriptor":
+                desc = dict(doc)
+                desc["configuration"] = dict(desc["configuration"])
+                for k, v in desc["configuration"].items():
+                    # try not to mutate the input!
+                    desc["configuration"][k] = dict(v)
+                    desc["configuration"][k]["timestamps"] = {
+                        k: t - time_offset for k, t in v["timestamps"].items()
+                    }
+                desc_bundle = desc_map[desc["uid"]] = run_bundle.compose_descriptor(
+                    name=desc["name"],
+                    time=desc["time"] - time_offset,
+                    data_keys=desc["data_keys"],
+                    configuration=desc["configuration"],
+                    object_keys=desc["object_keys"],
+                )
+                yield "descriptor", desc_bundle.descriptor_doc
+            elif name == "event_page":
+                desc_bundle = desc_map[doc["descriptor"]]
+                event_page = desc_bundle.compose_event_page(
+                    seq_num=doc["seq_num"],
+                    data=doc["data"],
+                    time=[t - time_offset for t in doc["time"]],
+                    timestamps={
+                        k: [t - time_offset for t in v]
+                        for k, v in doc["timestamps"].items()
+                    },
+                    filled=doc["filled"],
+                )
+                yield "event_page", event_page
+            elif name == "stop":
+                yield name, run_bundle.compose_stop(
+                    exit_status=doc["exit_status"],
+                    time=doc["time"] - time_offset,
+                    reason=doc.get("reason", ""),
+                )
+            else:
+                raise Exception("unexpected document!")