diff --git a/scripts/releases.py b/scripts/releases.py
new file mode 100644
index 00000000..27f3d0a8
--- /dev/null
+++ b/scripts/releases.py
@@ -0,0 +1,108 @@
+import argparse
+import logging
+import os.path
+import pandas as pd
+
+
+def create_release_stats(studies_file, release=None, date=None, size=None):
+    studies = pd.read_csv(studies_file, delimiter='\t', encoding='utf-8')
+
+    df = pd.DataFrame(columns=(
+        "Date",
+        "Data release",
+        "Code version",
+        "Sets",
+        "Wells",
+        "Experiments",
+        "Images",
+        "Planes",
+        "Size (TB)",
+        "Files (Million)",
+        "DB Size (GB)",
+    ))
+
+    if not release:
+        release = max(studies['Introduced'])
+    index = studies['Introduced'] <= release
+    if not date:
+        date = "TBD"
+    if not size:
+        size = "TBD"
+    df.loc[0] = (
+        date,
+        release,
+        get_release_code(release),
+        int(studies[index]['Sets'].sum()),
+        int(studies[index]['Wells'].sum()),
+        "",
+        int(studies[index]['5D Images'].sum()),
+        int(studies[index]['Planes'].sum()),
+        studies[index]['Size (TB)'].sum(),
+        studies[index]['# of Files'].sum() / 10 ** 6,
+        size)
+    return(df)
+
+
+def print_release_stats(df, fmt, target=None):
+    # fmt can be any of the pandas.Dataframe.to_{printfmt} methods
+    if fmt == 'tsv':
+        if target:
+            out = df.to_csv(
+                target, sep='\t', mode='a', header=False, index=False)
+            return
+        else:
+            out = df.to_csv(sep='\t', header=False, index=False)
+    elif fmt in ('json',):
+        out = getattr(df, f'to_{fmt}')()
+    else:
+        out = getattr(df, f'to_{fmt}')(index=False)
+    print(out)
+
+
+def get_release_code(relase_name):
+    number = relase_name[len('prod'):]
+    patch = int(number) % 10
+    minor = int((int(number) - patch) / 10)
+    return "0.%s.%s" % (minor, patch)
+
+
+def main():
+    parser = argparse.ArgumentParser("Generate release statistics")
+    parser.add_argument(
+        "--release", default=None, help="Name of the release")
+    parser.add_argument(
+        "--release-date", default=None, help="Date of the release")
+    parser.add_argument(
+        "--db-size", default=None,
+        help="Size of the database for the release in GB")
+    parser.add_argument("--format", default="tsv", help=(
+        "Output format, includes 'string', 'csv', 'tsv' (default), and "
+        "'json'. "
+        "'string' is the most human-readable (fixed width columns). "
+        "If tsv is selected and a file called releases.tsv exists in the "
+        "same directory as the file specified by studies_file, the output "
+        "will be appended to this file."
+    ))
+    parser.add_argument('-v', '--verbose', action='count', default=0)
+    parser.add_argument(
+        "studies_file", help="Path to TSV file containing study stats")
+    ns = parser.parse_args()
+
+    levels = [logging.WARNING, logging.INFO, logging.DEBUG]
+    level = levels[min(len(levels)-1, ns.verbose)]
+    logging.basicConfig(
+        level=level, format="%(asctime)s %(levelname)s %(message)s")
+
+    df = create_release_stats(
+        ns.studies_file, release=ns.release, date=ns.release_date,
+        size=ns.db_size)
+    releases_file = os.path.join(
+        os.path.dirname(ns.studies_file), 'releases.tsv')
+    if os.path.exists(releases_file):
+        print_release_stats(df, ns.format, target=releases_file)
+    else:
+        print_release_stats(df, ns.format)
+
+
+if __name__ == "__main__":
+    main()