How to Install Anaconda on Windows: Python and Jupyter Notebook Setup

To install Anaconda on Windows, download the 64-bit graphical installer from Anaconda's download page, run it for Just Me, leave the PATH checkbox unticked, and then work from Anaconda Prompt. Anaconda bundles Python with the scientific stack and the conda environment manager, which makes it the path of least resistance for data work on Windows.

Most broken installs trace back to two moments: ticking "Add Anaconda3 to my PATH", which Anaconda itself advises against, and installing project packages into base. This guide is for Windows users setting up Python for data science, and it walks through each installer screen, a clean project environment, a working Jupyter kernel and the errors you are most likely to see.

This guide follows Anaconda's Windows installation instructions as of September 2026, when the latest release was Anaconda Distribution 2026.07. Mac users should follow the macOS version of this guide instead, because the installer and shell setup differ.

Before you install Anaconda on Windows

Check three things first:

• Windows version. Anaconda's system requirements list 64-bit Windows 11.
• Disk space. Anaconda's Anaconda-or-Miniconda comparison gives about 9.7 GB of install space for Anaconda Distribution and about 900 MB for Miniconda.
• Licence terms. Under Anaconda's terms of service, use is free for individuals, eligible academic and non-profit institutions, and for-profit organisations with 200 or fewer employees and contractors. Larger for-profit organisations need a paid business licence. If that applies to your employer, check with IT before you install; Anaconda also states that conda and conda-forge are free for anyone.

How do you install Anaconda on Windows step by step?

1. Download. Go to anaconda.com/download and choose the Windows 64-bit graphical installer. Registration is optional; the page offers a "Skip registration" link.
2. Run the installer. Open it from your Downloads folder, select Next, and agree to the terms of service.
3. Choose Just Me. Anaconda recommends Just Me for the current user. All Users needs administrator rights and suits shared machines.
4. Choose the install location. The default under your user folder is fine. However, Anaconda recommends a folder with no spaces or special characters, so avoid paths such as C:\Program Files.
5. Review the advanced options. This is the step that matters:
   • Create shortcuts is ticked by default; leave it on.
   • Add Anaconda3 to my PATH environment variable: leave it unchecked. Anaconda does not recommend it, because it permanently puts Anaconda's Python ahead of other installations.
   • Register Anaconda3 as my default Python (the label includes the Python version): generally fine, since it lets VS Code and PyCharm find the interpreter.
   • Clear the package cache upon completion: optional; it runs conda clean --all --force-pkgs-dirs to recover disk space.
6. Install, then select Next twice and Finish.

How do you check the installation worked?

Open Anaconda Prompt from the Start menu, not the regular Command Prompt. Anaconda Prompt activates conda for that session, which is exactly why you did not need the PATH option. The prompt should start with (base):

(base) C:\Users\YourName>

Then run:

conda --version
python --version
conda info

If those respond, the install worked.

How do you create a conda environment on Windows?

Run conda create --name <project> python=3.12 in Anaconda Prompt, then conda activate <project>. The most important habit: do not install project packages into base. A broken base environment is awkward to repair and can mean reinstalling Anaconda.

conda create --name myproject python=3.12
conda activate myproject

The first time you use Anaconda's default channels, conda may ask you to accept their terms of service. Answer a to accept, or run conda tos accept once; Anaconda's ToS command-line guide explains the prompt.

Your prompt then changes to show (myproject). Now install what you need:

conda install pandas numpy matplotlib scikit-learn jupyter

Useful environment commands:

conda env list                      # all environments
conda deactivate                    # back to base
conda env remove --name myproject   # delete it
conda list                          # packages in the active env

To share an environment:

conda env export --from-history > environment.yml
conda env create -f environment.yml

A plain conda env export pins Windows-specific builds, so the file often fails on macOS or Linux. By contrast, --from-history records only the packages you explicitly asked for, which travels across operating systems.

How do you add a conda environment to Jupyter?

Install ipykernel in the environment and register it as a kernel. A new environment will not appear in Jupyter until you do:

conda activate myproject
conda install ipykernel
python -m ipykernel install --user --name myproject --display-name "Python (myproject)"

This is the single most common point of confusion. The packages installed fine, but the notebook is running a different environment and therefore cannot see them.

Launch Jupyter

Activate the environment, change to your project folder and run jupyter lab:

conda activate myproject
cd C:\Users\YourName\projects\analysis
jupyter lab

A browser then opens at localhost:8888. Start Jupyter from your project folder, because the file browser is rooted wherever you launched it.

jupyter lab is the current interface, while jupyter notebook still works if you prefer the classic one; jupyter.org's install page lists both. Stop the server with Ctrl+C twice in the terminal, since closing the browser tab leaves it running.

Quick reference: Anaconda commands on Windows

Common problems after you install Anaconda on Windows

"conda is not recognized as an internal or external command" You are in the regular Command Prompt. Use Anaconda Prompt instead. To enable conda in PowerShell permanently, run conda init powershell from Anaconda Prompt, then reopen PowerShell.

Environment activates but packages are missing You probably installed into base and then activated the project environment. Check with conda list after activating, and reinstall inside the active environment.

Solving environment takes forever Since conda 23.10, the much faster libmamba solver has been the default, according to the conda-libmamba-solver guide. If you are on an older conda, update it with conda update -n base conda, or switch explicitly:

conda install -n base conda-libmamba-solver
conda config --set solver libmamba

Kernel keeps dying in Jupyter It is usually memory. Check whether you are loading a file larger than available RAM, and read it in chunks instead; the iterators and generators guide shows the pattern.

Package not available on conda Try conda-forge first, then pip:

conda install -c conda-forge package-name
pip install package-name

Install conda packages before pip ones in a given environment, because mixing them in the other order is where dependency conflicts usually come from.

Is Miniconda a better choice?

If about 9.7 GB is too much, install Miniconda from the same download page. It gives you the same conda tooling with only Python and its dependencies, and you then add what you need:

conda create -n ds python=3.12 pandas numpy matplotlib scikit-learn jupyter

Miniconda is faster to install, uses less disk, and you always know what is in your environment. The licence terms above apply to both when you use Anaconda's default channels.

Related reading

The macOS version of this guide covers Apple Silicon. Once set up, start with Python's building blocks or the NumPy tutorial series. The planned Anaconda installation guide will also cover Linux.