diff --git a/src/lerner_lab_to_nwb/seiler_2024/seiler_2024_convert_dataset.py b/src/lerner_lab_to_nwb/seiler_2024/seiler_2024_convert_dataset.py
index a38e1ca..21049a6 100644
--- a/src/lerner_lab_to_nwb/seiler_2024/seiler_2024_convert_dataset.py
+++ b/src/lerner_lab_to_nwb/seiler_2024/seiler_2024_convert_dataset.py
@@ -52,6 +52,7 @@
     split_western_blot,
 )
 from lerner_lab_to_nwb.seiler_2024.medpc_helpers import get_medpc_variables, read_medpc_file
+from lerner_lab_to_nwb.seiler_2024.seiler_2024csvbehaviorinterface import Seiler2024CSVBehaviorInterface
 
 
 def dataset_to_nwb(
@@ -252,6 +253,32 @@ def get_session_key_from_kwargs(session_to_nwb_kwargs: dict):
     return session_key
 
 
+def get_subject_session_start(session_to_nwb_kwargs: dict) -> tuple[str, str, str]:
+    """Get the subject ID, start date, and start time of a session.
+
+    For CSV behavior files, the start date and time are read from the CSV itself.
+
+    Parameters
+    ----------
+    session_to_nwb_kwargs : dict
+        The arguments for session_to_nwb.
+
+    Returns
+    -------
+    tuple[str, str, str]
+        The subject ID, start date (mm/dd/yy), and start time (HH:MM:SS) of the session.
+    """
+    behavior_file_path = session_to_nwb_kwargs["behavior_file_path"]
+    if behavior_file_path.suffix in (".csv", ".CSV"):
+        behavior_metadata = Seiler2024CSVBehaviorInterface(file_path=str(behavior_file_path)).get_metadata()["Behavior"]
+        start_date = behavior_metadata["start_date"]
+        start_time = behavior_metadata["start_time"]
+    else:
+        start_date = session_to_nwb_kwargs["session_conditions"]["Start Date"]
+        start_time = session_to_nwb_kwargs["session_conditions"]["Start Time"]
+    return session_to_nwb_kwargs["subject_id"], start_date, start_time
+
+
 def safe_session_to_nwb(*, session_to_nwb_kwargs: dict, exception_file_path: Union[Path, str]):
     """Convert a session to NWB while handling any errors by recording error messages to the exception_file_path.
 
@@ -731,6 +758,10 @@ def opto_to_nwb(
                             continue
                         unique_session_keys.add(session_key)
                         session_to_nwb_args_per_session.append(session_to_nwb_args)
+    subject_session_starts = {
+        get_subject_session_start(session_to_nwb_args) for session_to_nwb_args in session_to_nwb_args_per_session
+    }
+
     # DLS Excitatory raw files by date
     raw_files_by_date_path = data_dir_path / "Opto Experiments" / "DLS Excitatory"
     start_dates, start_times, msns, file_paths, subjects, box_numbers = [], [], [], [], [], []
@@ -765,6 +796,8 @@ def opto_to_nwb(
         }
         if subject in partial_subject_ids_to_subject_id:
             subject = partial_subject_ids_to_subject_id[subject]
+        if (subject, start_date, start_time) in subject_session_starts:
+            continue  # already converted from the subject's own folder, which also gives the treatment
         session_to_nwb_args = dict(
             data_dir_path=data_dir_path,
             output_dir_path=output_dir_path,
diff --git a/tests/test_opto_session_deduplication.py b/tests/test_opto_session_deduplication.py
new file mode 100644
index 0000000..8e1d93b
--- /dev/null
+++ b/tests/test_opto_session_deduplication.py
@@ -0,0 +1,70 @@
+from pathlib import Path
+
+import pytest
+
+from lerner_lab_to_nwb.seiler_2024.seiler_2024_convert_dataset import opto_to_nwb
+
+TREATMENT_FOLDERS = [
+    "DLS Excitatory/ChR2",
+    "DLS Excitatory/EYFP",
+    "DLS Excitatory/Scrambled",
+    "DMS Excitatory/ChR2",
+    "DMS Excitatory/EYFP",
+    "DMS Excitatory/Scrambled",
+    "DMS Inhibitory/Group 1/Halo",
+    "DMS Inhibitory/Group 1/EYFP",
+    "DMS Inhibitory/Group 1/Scrambled",
+    "DMS Inhibitory/Group 2/NpHr",
+    "DMS Inhibitory/Group 2/EYFP",
+    "DMS Inhibitory/Group 2/Scrambled",
+]
+
+
+def medpc_run(subject: str, box: str) -> str:
+    return (
+        "Start Date: 09/29/20\n"
+        "End Date: 09/29/20\n"
+        f"Subject: {subject}\n"
+        "Experiment: \n"
+        "Group: \n"
+        f"Box: {box}\n"
+        "Start Time: 11:36:53\n"
+        "End Time: 12:36:45\n"
+        "MSN: Footshock Degradation Left\n"
+    )
+
+
+@pytest.fixture
+def data_dir_path(tmp_path) -> Path:
+    opto_path = tmp_path / "Opto Experiments"
+    for treatment_folder in TREATMENT_FOLDERS:
+        (opto_path / treatment_folder).mkdir(parents=True)
+
+    # 300.405's run exported to its own folder as a CSV
+    subject_folder_path = opto_path / "DLS Excitatory" / "Scrambled" / "300_405"
+    subject_folder_path.mkdir()
+    (subject_folder_path / "300.405_09-29-20.csv").write_text(
+        ",Start Date,End Date,Subject,Experiment,Group,Box,Start Time,End Time,MSN\n"
+        "0,09/29/20,09/29/20,300.405,,,5,11:36:53,12:36:45,Footshock Degradation Left\n"
+    )
+
+    # The same run and another mouse's run in the date-organized MED-PC file
+    (opto_path / "DLS Excitatory" / "2020-09-29").write_text(
+        "\n" + medpc_run(subject="418.404", box="1") + "\n" + medpc_run(subject="300.405", box="5")
+    )
+    return tmp_path
+
+
+class TestOptoToNwb:
+    def test_by_date_run_already_in_subject_folder_is_skipped(self, data_dir_path, tmp_path):
+        session_to_nwb_args_per_session = opto_to_nwb(
+            data_dir_path=data_dir_path,
+            output_dir_path=tmp_path / "nwb",
+            start_variable="Start Date",
+            verbose=False,
+        )
+        subject_and_treatment = sorted(
+            (session_to_nwb_args["subject_id"], session_to_nwb_args["optogenetic_treatment"])
+            for session_to_nwb_args in session_to_nwb_args_per_session
+        )
+        assert subject_and_treatment == [("300.405", "ChR2Scrambled"), ("418.404", "Unknown")]