Fix "TypeError: 'NoneType' object is not subscriptable" (and Other NoneType Errors) in Python
Why Python says 'NoneType' object is not subscriptable, has no attribute, or is not iterable — the functions that silently return None, how to find the culprit, and how to fix it properly.
By Software Side Hustles Editorial Team · Published · Updated · 4 min read
Quick answer: a value you expected to be a list, dict, string or object is actually None. 'NoneType' object is not subscriptable means you used [...] on it; has no attribute 'x' means you used .x; is not iterable means you looped over it. Find which variable is None (the traceback's last line in your code), then find where it was assigned — usually a function with no return, a method like list.sort() that returns None, or a lookup that found nothing.
The NoneType error family
| Error | You did this to None |
|---|---|
TypeError: 'NoneType' object is not subscriptable | x[0], x["key"] |
AttributeError: 'NoneType' object has no attribute 'group' | x.group(), x.text |
TypeError: 'NoneType' object is not iterable | for i in x, list(x) |
TypeError: cannot unpack non-iterable NoneType object | a, b = x |
TypeError: unsupported operand type(s) for +: 'NoneType' and 'int' | x + 1 |
They are all the same bug: something returned None and the code carried on as if it hadn't.
Cause 1: in-place methods return None
scores = [70, 95, 82]top = scores.sort(reverse=True)[0]# TypeError: 'NoneType' object is not subscriptablelist.sort(), list.append(), list.extend(), list.reverse(), dict.update(), random.shuffle() and set.add() all modify the object in place and return None. Python does this deliberately, so you cannot mistake an in-place change for a copy. Either call the method on its own line, or use the function that returns a new object:
scores.sort(reverse=True)top = scores[0]# ortop = sorted(scores, reverse=True)[0]The classic variant is items = items.append(x), which replaces your list with None.
Cause 2: a function with no return statement
def load_config(path): with open(path) as f: config = json.load(f) # forgot: return config
settings = load_config("app.json")settings["debug"] # 'NoneType' object is not subscriptableA Python function without return — or that reaches the end of a branch without one — returns None. Look out for functions where only some branches return:
def find_user(users, email): for u in users: if u["email"] == email: return u # falls through: returns None when not foundThat may be correct behaviour, but then every caller must handle None. Type hints make this explicit — -> dict | None — and a type checker such as mypy or Pyright will flag callers that forget.
Cause 3: lookups that find nothing
Many APIs signal "not found" by returning None instead of raising:
match = re.match(r"(\d+)-(\d+)", "abc")match.group(1) # AttributeError: 'NoneType' object has no attribute 'group'| Call | Returns None when |
|---|---|
re.match, re.search, re.fullmatch | the pattern does not match |
dict.get(key) | the key is missing |
os.environ.get("VAR") | the variable is not set |
BeautifulSoup soup.find(...) | no element matches |
SQLAlchemy query.first(), session.get(...) | no row matches |
cursor.fetchone() | no rows remain |
Check before you use the result:
match = re.match(r"(\d+)-(\d+)", text)if match is None: raise ValueError(f"Expected a range like 10-20, got {text!r}")start, end = match.groups()From Python 3.8 the walrus operator keeps it compact: if (m := re.match(pattern, text)):.
Cause 4: values that are None by design
Optional database columns, JSON null, optional function arguments with a None default — these flow into code that assumes a value. Decide at the boundary: replace with a default (name = row["name"] or "Anonymous"), reject the input, or carry the None deliberately with a type hint saying so.
Careful with or for defaults: count = row["count"] or 10 turns a legitimate 0 into 10. Use 10 if row["count"] is None else row["count"] when zero is valid.
How to find which value is None
- Read the last frame of the traceback that is in your code. It shows the exact line. See how to read a stack trace.
- If the line has several candidates —
data["user"]["orders"][0]— split it or print each part:print(type(data), type(data.get("user"))). - Use the debugger:
breakpoint()on the line before, then inspect variables. In VS Code or PyCharm, enable "break on raised exceptions". - Python 3.11+ highlights the failing sub-expression with
^^^^markers under the line in the traceback, which often answers the question immediately. - Trace the variable back to its assignment and ask: which call could return
Nonehere?
Preventing NoneType errors
- Use type hints with
Optional[...]/X | Noneand run a type checker in CI. - Return early with clear errors rather than passing
Nonealong. - Prefer exceptions for "should never happen" —
d[key]raisingKeyErroris often more helpful thand.get(key)returningNonethree function calls before the crash. - Avoid reassigning results of in-place methods. If a linter is available, Pylint flags
x = x.sort()("assignment-from-none").
Frequently asked questions
What does "subscriptable" mean?
An object is subscriptable if it supports the [] operator — it implements __getitem__. Lists, tuples, strings and dicts are; None, integers and functions are not.
Why does my function return None when I printed the value?
Printing is not returning. print(result) shows the value on screen; the caller still receives None unless you write return result.
Is if x: a good None check?
Only if empty strings, zero and empty lists should be treated the same as None. Otherwise write if x is not None:.
Why is None instead of == None?
None is a singleton, so identity is the precise check, and == can be overridden by classes. See equality operators explained.
Where this skill pays off
The x = x.sort() bug shows up in AI-written Python more often than you would expect, and spotting it instantly is part of what makes a strong code evaluator. If you can explain bugs like this clearly, read how software engineers earn money training AI models.