IRR and NPV are built from the same cash flows and usually agree on whether a single project is worth doing — but when comparing two mutually exclusive projects, they can rank them in opposite order, and knowing which to trust matters.
What each one actually measures
NPV answers "how much value, in today's dollars, does this project create?" — an absolute dollar figure.
IRR answers "what annualized rate of return does this project generate?" — a percentage, independent of project size.
Why they can disagree
IRR is scale-blind. A project requiring $10,000 that returns 50% IRR generates far less total value than a project requiring $1,000,000 that returns 20% IRR — but IRR alone would rank the smaller project "better." NPV, by contrast, captures the actual dollar value created, which is what a business ultimately cares about.
A worked comparison
| Project | Investment | IRR | NPV (at 10% discount rate) |
|---|---|---|---|
| A | $10,000 | 45% | $4,200 |
| B | $500,000 | 22% | $85,000 |
Project A "wins" on IRR. Project B creates 20× more actual value. If you can only choose one and have the capital available, Project B is very likely the better choice despite the lower IRR — this is exactly the trap IRR-only decision-making falls into.
When IRR is still the right primary metric
- Capital-constrained screening: when you're choosing among many small projects and capital (not a single go/no-go decision) is the binding constraint, IRR-based ranking to maximize return per dollar deployed can be the right approach
- Communicating with non-technical stakeholders: a percentage is often more intuitive to discuss than a dollar NPV figure, especially for comparing to a hurdle rate
The standard recommendation
Use NPV as the primary decision criterion when comparing mutually exclusive projects of different sizes — it directly measures value created. Use IRR as a secondary sanity check and for communicating expected annualized return, but don't let it override NPV when the two disagree on ranking.
Frequently asked questions
What if a project has multiple IRRs? This happens when cash flows change sign more than once (e.g., a large cost partway through the project). NPV remains well-defined in this case, which is another reason to default to NPV when the two conflict or when IRR looks unstable.
Does MIRR fix the disagreement between IRR and NPV? MIRR fixes IRR's unrealistic reinvestment assumption but doesn't fix the scale-blindness problem — NPV is still the metric to trust for ranking projects of different sizes.
Use the IRR Calculator alongside your own NPV calculation to see both metrics for the same cash flow series before deciding between projects.