Home / Scenarios / Python / Anaconda "DLL load failed"

🐍

Developers / data scientists

Python / Anaconda "DLL load failed"

After pip install, importing a package raises "DLL load failed while importing xxx"

Typical symptoms

ImportError: DLL load failed while importing _ssl
ImportError: DLL load failed: The specified module could not be found
numpy / scipy / torch / tensorflow import errors
Python crashing immediately after activating a Conda env

Official safe fix steps

Install the latest VC++ 2015-2022 Redistributable

Python 3.5+ is built with VS 2015+, and every C-extension package (numpy, scipy, torch, pillow, psycopg2, etc.) depends on vcruntime140. x64 Python needs x64 redist, x86 Python needs x86 redist.
Make sure Python bitness matches your wheels

Run python -c "import struct; print(struct.calcsize('P')*8)" in PowerShell to see if your Python is 32 or 64 bit. Then make sure the wheels (.whl) you pip-install match.
```
python -c "import struct; print(struct.calcsize('P') * 8)"
```
Recreate the conda env instead of repairing it

When a conda env breaks, "conda create -n new_env --clone broken_env" then deleting the old one is far faster than patching DLLs by hand.
Capture load failures with Process Monitor

Advanced: SysInternals Process Monitor filtered on python.exe with "Result is NOT SUCCESS" shows exactly which DLL load step failed — far more useful than the error message.

DLLs commonly seen in this scenario

vcruntime140.dll Visual C++ 2015-2022 Redistributable vcruntime140_1.dll Visual C++ 2015-2022 Redistributable msvcp140.dll Visual C++ 2015-2022 Redistributable msvcp140_1.dll Visual C++ 2015-2022 Redistributable ucrtbase.dll Universal C Runtime

💡 Still stuck? Check the general Fix Guide or look up your exact error code at Error Codes.

← Back to scenarios