os.path.join() in Python: Join Paths the Right Way

os.path.join() in Python combines folder and file names into one path, using the right separator for your operating system: a backslash on Windows and a forward slash on macOS and Linux.

import os
os.path.join("reports", "2026", "sales.csv")
# Windows: reports\2026\sales.csv    Mac/Linux: reports/2026/sales.csv

My machine runs Python 3.12.5 on Windows, so the screenshots show backslashes; I also use posixpath to show what macOS and Linux would print.

I’ll cover the basics, Windows vs Linux output, the absolute-path rule that confuses everyone, joining a list, the pathlib alternative, and building paths relative to your script.

How os.path.join works in Python

Pass any number of strings and it puts one separator between each. It won’t double a slash that’s already there:

import os

path = os.path.join("reports", "2026", "october", "sales.csv")
print(path)
print("separator on this OS:", repr(os.sep))

folder = "invoices"
print(os.path.join(folder, "INV-1042.pdf"))
print(os.path.join("data/", "users.json"))       # an existing trailing slash isn't doubled

Output:

reports\2026\october\sales.csv
separator on this OS: '\\'
invoices\INV-1042.pdf
data/users.json

Why not just glue strings with + "/" +? Because then you have to remember separators and trailing slashes yourself, and your code breaks the moment a path ends with one.

os.path.join on Windows vs Linux and macOS

os.path is really ntpath on Windows and posixpath everywhere else. Importing both lets you see each platform’s result from one machine:

import ntpath, posixpath

parts = ("home", "jane", "projects", "app.py")
print("Windows rules:", ntpath.join(*parts))
print("Mac/Linux:    ", posixpath.join(*parts))

print(ntpath.join("C:\\Users", "Public", "notes.txt"))
print(posixpath.join("/var/log", "nginx", "access.log"))

Output:

Windows rules: home\jane\projects\app.py
Mac/Linux:     home/jane/projects/app.py
C:\Users\Public\notes.txt
/var/log/nginx/access.log

Why does os.path.join ignore my first path?

This is the big gotcha. If any part is an absolute path, everything before it is thrown away. A leading slash on the second argument is usually an accident:

import ntpath, posixpath

print(posixpath.join("/srv/app", "/etc/config.ini"))      # absolute part wins
print(posixpath.join("/srv/app", "etc/config.ini"))       # relative part is appended

print(ntpath.join("C:\\app", "D:\\data", "file.csv"))      # new drive resets everything
print(ntpath.join("C:\\app", "\\data", "file.csv"))        # rooted: keeps the drive C:
print(ntpath.join("C:\\app", "logs", ""))                  # empty last part adds a separator

Output:

/etc/config.ini
/srv/app/etc/config.ini
D:\data\file.csv
C:\data\file.csv
C:\app\logs\

The fix is to strip the leading slash, for example part.lstrip("/\\"), when the piece comes from user input or a URL. And an empty last argument adds a trailing separator, which is handy for folder prefixes.

Joining a list of path parts

os.path.join() takes separate arguments, not a list, so unpack the list with *. In a loop, it’s an easy way to build one path per item:

import os

parts = ["exports", "2026", "q4", "summary.csv"]
print(os.path.join(*parts))                   # unpack a list

base = os.path.join("backups", "stores")
for city in ["austin", "denver", "boston"]:
    print(os.path.join(base, city, "orders.db"))

try:
    os.path.join("reports", 2026)              # every part must be a string or path
except TypeError as err:
    print("TypeError:", err)

Output:

exports\2026\q4\summary.csv
backups\stores\austin\orders.db
backups\stores\denver\orders.db
backups\stores\boston\orders.db
TypeError: join() argument must be str, bytes, or os.PathLike object, not 'int'

Each part must be a string or a path object. Convert numbers first with str(); there’s more on that in converting an int to a string.

os.path.join vs pathlib

pathlib does the same job with the / operator and gives you useful extras like .name, .suffix and .parent:

import os
from pathlib import Path

old = os.path.join("reports", "2026", "sales.csv")
new = Path("reports") / "2026" / "sales.csv"

print(old)
print(new)
print(str(new) == old)
print(new.name, new.suffix, new.parent)          # extra helpers pathlib gives you
print(Path("/srv/app") / "/etc/config.ini")      # same absolute-path rule

Output:

reports\2026\sales.csv
reports\2026\sales.csv
True
sales.csv .csv reports\2026
\etc\config.ini

I use pathlib for new code and os.path.join when I’m working in older code that already uses os. Both are fine. For creating folders with either, see how to make a directory in Python.

Building paths relative to your script

Relative paths depend on the folder you run Python from, which is a common source of FileNotFoundError. Anchor them to the script’s own folder instead:

import os

script_dir = os.path.dirname(os.path.abspath(__file__))
data_file = os.path.join(script_dir, "data", "prices.csv")

print("ends with:", os.path.join(*data_file.split(os.sep)[-2:]))
print("normalized:", os.path.normpath(os.path.join("reports", "..", "exports", ".", "q4")))
print("exists?", os.path.exists(data_file))

Output:

ends with: data\prices.csv
normalized: exports\q4
exists? False

os.path.normpath() cleans up .. and . parts. If you’d rather change folders, see getting the current directory in Python.

Here are more file path guides to try:

• Get the filename from a path
• Get a file extension in Python
• Check if a file exists in Python
• Python with open() as file

Frequently asked questions