diff --git a/src/contributor_stats/main.py b/src/contributor_stats/main.py
index c5bd9d8..1ba62b0 100644
--- a/src/contributor_stats/main.py
+++ b/src/contributor_stats/main.py
@@ -1,10 +1,11 @@
 import sys
 import os
+import subprocess
 import unicodedata
 import logging
 from pathlib import Path
-from typing import Dict, Tuple, Any, MutableMapping
-from collections import OrderedDict
+from typing import Any, Dict, Iterable, List, MutableMapping, Optional, Set, Tuple
+from collections import OrderedDict, defaultdict
 from contextlib import contextmanager
 
 original_cwd = os.getcwd()
@@ -17,9 +18,20 @@
 
 AuthorInfo = MutableMapping[str, dict]
 Table = MutableMapping[str, AuthorInfo]
+# Author name -> commit email -> number of commits made with it
+Emails = MutableMapping[str, MutableMapping[str, int]]
 
 zero_row = OrderedDict(commits=0, active_days=[], lines_added=0, lines_removed=0, blame=0)
 
+# Emails that identify a git client default or a shared/anonymous address rather
+# than a person, and must therefore never be used to merge two authors together.
+GENERIC_EMAILS = {
+    "noreply@github.com",
+    "you@example.com",
+    "root@localhost",
+    "git@localhost",
+}
+
 
 def foldername(path) -> str:
     if os.path.isdir(path):
@@ -39,25 +51,159 @@ def merge_author(a1: MutableMapping, a2: MutableMapping) -> AuthorInfo:
     return a1
 
 
-def get_authorInfos(data) -> AuthorInfo:
+def normalize_name(name: str) -> str:
+    """
+    Normalize an author name, so that different spellings of the same name compare equal.
+    """
+    # Run the following and be amazed by the power of Unicode:
+    #   bool('å' == "å")  # False
+    # This weird unicode char was in Måns name, so now we have to unicode normalize everything.
+    # Never done this before, so thanks for making me learn Måns, or perhaps should I write Måns.
+    new_name = unicodedata.normalize("NFKC", name)
+    if new_name != name:
+        logger.info("Name '{}' was normalized to '{}'".format(name, new_name))
+        name = new_name
+
+    return name.replace("å", "å")
+
+
+def normalize_email(email: str) -> str:
+    """
+    Normalize a commit email, returning an empty string for the emails that can't be
+    used to identify an author (malformed, or a well-known generic address).
+    """
+    email = email.strip().lower()
+    if "@" not in email or email in GENERIC_EMAILS:
+        return ""
+    return email
+
+
+def git_author_emails(path) -> Emails:
+    """
+    Maps each author name in the history to the emails they have committed with,
+    and how many commits they made with each of them.
+
+    Uses the same (mailmap-resolved) names as gitstats, so that the keys line up
+    with the names in the tables.
+    """
+    output = subprocess.check_output(
+        ["git", "-C", str(path), "log", "--format=%aN%x09%aE", "HEAD"], text=True
+    )
+    emails: Emails = defaultdict(lambda: defaultdict(int))
+    for line in output.splitlines():
+        name, _, email = line.partition("\t")
+        email = normalize_email(email)
+        if not name or not email:
+            continue
+        emails[normalize_name(name)][email] += 1
+    return emails
+
+
+def merge_emails(emails: Iterable[Emails]) -> Emails:
+    """Combines the author -> email mappings of several repos into one."""
+    merged: Emails = defaultdict(lambda: defaultdict(int))
+    for repo_emails in emails:
+        for name, counts in repo_emails.items():
+            for email, commits in counts.items():
+                merged[name][email] += commits
+    return merged
+
+
+def group_names_by_email(emails: Emails) -> List[List[str]]:
+    """
+    Groups the author names that share at least one commit email.
+
+    Grouping is transitive: if "a" and "b" share an email, and "b" and "c" share
+    another, then all three end up in the same group.
+    """
+    parent = {name: name for name in emails}
+
+    def find(name: str) -> str:
+        while parent[name] != name:
+            parent[name] = parent[parent[name]]
+            name = parent[name]
+        return name
+
+    def union(name: str, other: str) -> None:
+        root, other_root = find(name), find(other)
+        if root != other_root:
+            parent[other_root] = root
+
+    names_by_email: Dict[str, List[str]] = defaultdict(list)
+    for name, counts in emails.items():
+        for email in counts:
+            names_by_email[email].append(name)
+
+    for group in names_by_email.values():
+        for other in group[1:]:
+            union(group[0], other)
+
+    groups: Dict[str, List[str]] = defaultdict(list)
+    for name in parent:
+        groups[find(name)].append(name)
+    return list(groups.values())
+
+
+def author_aliases(emails: Emails) -> Dict[str, str]:
+    """
+    Maps every author name that has an alias (another name used with one of the
+    same commit emails) to the single name to display them under.
+
+    The name with the most commits wins, ties broken by name to keep the result
+    deterministic (which also happens to prefer "Brayo" over "brayo", since
+    uppercase sorts first). The mapping is computed across all repos at once, so
+    that the same person gets the same name in every table.
+    """
+    aliases = {}
+    for group in group_names_by_email(emails):
+        if len(group) < 2:
+            continue
+        keep = min(group, key=lambda name: (-sum(emails[name].values()), name))
+        logger.info("Merging {} into '{}' (shared commit email)".format(
+            sorted(name for name in group if name != keep), keep))
+        for name in group:
+            if name != keep:
+                aliases[name] = keep
+    return aliases
+
+
+def get_authorInfos(data, aliases: Optional[Dict[str, str]] = None) -> Tuple[AuthorInfo, Dict[str, str]]:
+    """
+    Returns the stats per author, with the aliases of a person merged into a single
+    entry, and a mapping from every name in the history to the name it was merged
+    into (used to attribute blame lines to the merged author).
+    """
     names = data.getAuthors()
+    aliases = dict(aliases or {})
 
-    authorInfos = {}
+    authorInfos: AuthorInfo = {}
     for name in names:
         _authorInfo = data.getAuthorInfo(name)
-
-        # Run the following and be amazed by the power of Unicode:
-        #   bool('å' == "å")  # False
-        # This weird unicode char was in Måns name, so now we have to unicode normalize everything.
-        # Never done this before, so thanks for making me learn Måns, or perhaps should I write Måns.
-        _new_name = unicodedata.normalize("NFKC", name)
-        if _new_name != name:
-            logger.info("Name '{}' was normalized to '{}'".format(name, _new_name))
-            name = _new_name
-
-        name = name.replace("å", "å")
+        name = normalize_name(name)
+        if name in authorInfos:
+            # Two spellings of the name normalized into the same one
+            _authorInfo = merge_author(dict(authorInfos[name]), _authorInfo)
         authorInfos[name] = _authorInfo
 
+    # Every name maps to itself, until merged into another one
+    merged_into = {name: name for name in authorInfos}
+
+    def merge_into(keep: str, alias: str) -> None:
+        merged: Any = authorInfos.pop(alias)
+        if keep in authorInfos:
+            merged = merge_author(dict(authorInfos[keep]), merged)
+        authorInfos[keep] = merged
+        for name, into in merged_into.items():
+            if into == alias:
+                merged_into[name] = keep
+
+    # Merge the names that committed with the same email
+    for alias in sorted(authorInfos):
+        keep = aliases.get(alias, alias)
+        if keep != alias:
+            merge_into(keep, alias)
+
+    # Manual merges, for the aliases that don't share a commit email
     author_merges = [
         ("Erik Bjäreholt", ["Erik Bjäreholt", "Erik Bjareholt"]),
         ("Johan Bjäreholt", ["johan-bjareholt"]),
@@ -69,16 +215,12 @@ def get_authorInfos(data) -> AuthorInfo:
         ("Otto-AA", ["A_A"]),
         ("Brayo", ["brayo"])
     ]
-    for name, aliases in author_merges:
-        for alias in aliases:
+    for name, _aliases in author_merges:
+        for alias in _aliases:
             if alias in authorInfos:
-                to_keep = authorInfos.pop(alias)
-                if name in authorInfos:
-                    to_merge_with = authorInfos.pop(name)
-                    to_keep = merge_author(to_merge_with, to_keep)
-                authorInfos[name] = to_keep
+                merge_into(name, alias)
 
-    return authorInfos
+    return authorInfos, merged_into
 
 
 def git_blame_stats(path) -> dict[str, int]:
@@ -86,7 +228,6 @@ def git_blame_stats(path) -> dict[str, int]:
     Uses the following command to get stats of lines last touched by each author:
         git ls-tree --name-only -z -r HEAD -- $1 | xargs -0 -n1 git blame --line-porcelain | grep "^author "|sort|uniq -c|sort -nr
     """
-    import subprocess
     os.chdir(path)
     output = subprocess.check_output(
         "git ls-tree --name-only -z -r HEAD -- . | xargs -0 -n1 git blame --line-porcelain | grep '^author '|sort|uniq -c|sort -nr", shell=True, text=True
@@ -98,8 +239,10 @@ def git_blame_stats(path) -> dict[str, int]:
     return {" ".join(line.split()[2:]): int(line.split()[0]) for line in output.split("\n") if line}
 
 
-def generate_from_repo(path: Path) -> Tuple[str, Table]:
+def generate_from_repo(path: Path, aliases: Optional[Dict[str, str]] = None) -> Tuple[str, Table]:
     path = Path(path).resolve()
+    if aliases is None:
+        aliases = author_aliases(git_author_emails(path))
 
     # TODO: Could use caching to speed up (not much point since it usually runs in CI)
     data = gitstats.GitDataCollector()
@@ -117,10 +260,17 @@ def generate_from_repo(path: Path) -> Tuple[str, Table]:
     print("Generated stats for: {}".format(data.projectname))
 
     rows = {}
-    authorInfos = get_authorInfos(data)
+    authorInfos, merged_into = get_authorInfos(data, aliases)
+
+    # Attribute the blame lines of every alias to the author it was merged into
+    blame_by_author: Dict[str, int] = defaultdict(int)
+    for blamed_name, lines in blame.items():
+        name = normalize_name(blamed_name)
+        blame_by_author[merged_into.get(name, name)] += lines
+
     for name, info in authorInfos.items():
         rows[name] = merge_author(zero_row.copy(), info)
-        rows[name]["blame"] = blame.get(name, 0)  # type: ignore[assignment]
+        rows[name]["blame"] = blame_by_author.get(name, 0)  # type: ignore[assignment]
 
     for name in rows:
         rows[name]["blame_percent"] = rows[name]["blame"] / blame_lines * 100  # type: ignore
@@ -242,8 +392,12 @@ def main():
 
     print("Found repos: {}".format([str(r) for r in repos]))
 
+    # Resolve aliases across all repos at once, so that a person gets the same
+    # name in every table (and thus a single row in the merged "total" table)
+    aliases = author_aliases(merge_emails(git_author_emails(path) for path in repos))
+
     for path in repos:
-        repo_name, rows = generate_from_repo(str(path))
+        repo_name, rows = generate_from_repo(str(path), aliases)
         tables[repo_name] = rows
 
     tables["total"] = merge_tables(tables)
diff --git a/tests/test_main.py b/tests/test_main.py
new file mode 100644
index 0000000..3ad53b8
--- /dev/null
+++ b/tests/test_main.py
@@ -0,0 +1,152 @@
+from contributor_stats.main import (
+    author_aliases,
+    get_authorInfos,
+    group_names_by_email,
+    merge_emails,
+    normalize_email,
+    normalize_name,
+)
+
+
+class FakeData:
+    """Stands in for gitstats.GitDataCollector, which needs an actual repo."""
+
+    def __init__(self, authors):
+        self.authors = authors
+
+    def getAuthors(self):
+        return list(self.authors)
+
+    def getAuthorInfo(self, author):
+        return dict(self.authors[author])
+
+
+def author(commits, active_days, lines_added=0, lines_removed=0):
+    return dict(
+        commits=commits,
+        active_days=active_days,
+        lines_added=lines_added,
+        lines_removed=lines_removed,
+    )
+
+
+def test_normalize_name():
+    # NFD and NFC spellings of the same name are the same author
+    assert normalize_name("Måns") == normalize_name("Måns")
+
+
+def test_normalize_email():
+    assert normalize_email(" Erik@Bjareho.lt ") == "erik@bjareho.lt"
+    # Not usable to identify an author
+    assert normalize_email("not an email") == ""
+    assert normalize_email("noreply@github.com") == ""
+
+
+def test_group_names_by_email():
+    groups = group_names_by_email(
+        {
+            "Brayo": {"vukubrian@gmail.com": 2, "brayo@laptop.local": 1},
+            "brayo": {"vukubrian@gmail.com": 1},
+            "Erik": {"erik@bjareho.lt": 1},
+        }
+    )
+    assert sorted(sorted(group) for group in groups) == [["Brayo", "brayo"], ["Erik"]]
+
+
+def test_group_names_by_email_is_transitive():
+    groups = group_names_by_email(
+        {
+            "a": {"one@example.com": 1},
+            "b": {"one@example.com": 1, "two@example.com": 1},
+            "c": {"two@example.com": 1},
+        }
+    )
+    assert [sorted(group) for group in groups] == [["a", "b", "c"]]
+
+
+def test_author_aliases_picks_the_name_with_most_commits():
+    aliases = author_aliases(
+        {"Brayo": {"vukubrian@gmail.com": 5}, "brayo": {"vukubrian@gmail.com": 1}}
+    )
+    assert aliases == {"brayo": "Brayo"}
+
+
+def test_author_aliases_ties_are_deterministic():
+    emails = {"Brayo": {"vukubrian@gmail.com": 1}, "brayo": {"vukubrian@gmail.com": 1}}
+    assert author_aliases(emails) == {"brayo": "Brayo"}
+    assert author_aliases(dict(reversed(list(emails.items())))) == {"brayo": "Brayo"}
+
+
+def test_author_aliases_keeps_authors_without_a_shared_email_apart():
+    assert author_aliases(
+        {"Brayo": {"vukubrian@gmail.com": 1}, "Erik": {"erik@bjareho.lt": 1}}
+    ) == {}
+
+
+def test_merge_emails():
+    merged = merge_emails(
+        [
+            {"Brayo": {"vukubrian@gmail.com": 2}},
+            {"Brayo": {"vukubrian@gmail.com": 3}, "brayo": {"vukubrian@gmail.com": 1}},
+        ]
+    )
+    assert merged == {
+        "Brayo": {"vukubrian@gmail.com": 5},
+        "brayo": {"vukubrian@gmail.com": 1},
+    }
+
+
+def test_get_authorInfos_merges_aliases():
+    data = FakeData(
+        {
+            "Brayo": author(commits=2, active_days=["2025-01-01", "2025-01-02"], lines_added=10),
+            "brayo": author(commits=1, active_days=["2025-01-02"], lines_added=5),
+            "Erik": author(commits=1, active_days=["2025-01-03"], lines_added=1),
+        }
+    )
+    authorInfos, merged_into = get_authorInfos(data, {"brayo": "Brayo"})
+
+    assert set(authorInfos) == {"Brayo", "Erik"}
+    assert authorInfos["Brayo"]["commits"] == 3
+    assert authorInfos["Brayo"]["lines_added"] == 15
+    assert authorInfos["Brayo"]["active_days"] == {"2025-01-01", "2025-01-02"}
+    # so that blame lines of an alias are attributed to the merged author
+    assert merged_into["brayo"] == "Brayo"
+    assert merged_into["Brayo"] == "Brayo"
+
+
+def test_get_authorInfos_merges_into_a_name_not_in_the_repo():
+    data = FakeData({"brayo": author(commits=1, active_days=["2025-01-01"])})
+    authorInfos, merged_into = get_authorInfos(data, {"brayo": "Brayo"})
+
+    assert set(authorInfos) == {"Brayo"}
+    assert authorInfos["Brayo"]["commits"] == 1
+    assert merged_into["brayo"] == "Brayo"
+
+
+def test_get_authorInfos_still_applies_the_manual_merges():
+    # These aliases don't share a commit email, so they need the manual list
+    data = FakeData(
+        {
+            "dependabot[bot]": author(commits=1, active_days=["2025-01-01"]),
+            "dependabot-preview[bot]": author(commits=2, active_days=["2025-01-02"]),
+        }
+    )
+    authorInfos, merged_into = get_authorInfos(data)
+
+    assert set(authorInfos) == {"dependabot[bot]"}
+    assert authorInfos["dependabot[bot]"]["commits"] == 3
+    assert merged_into["dependabot-preview[bot]"] == "dependabot[bot]"
+
+
+def test_get_authorInfos_merges_names_that_normalize_to_the_same_one():
+    data = FakeData(
+        {
+            "Måns": author(commits=1, active_days=["2025-01-01"]),
+            "Måns": author(commits=2, active_days=["2025-01-02"]),
+        }
+    )
+    authorInfos, _ = get_authorInfos(data)
+
+    assert len(authorInfos) == 1
+    assert list(authorInfos.values())[0]["commits"] == 3