AACE Estimate Classes Explained

What Class 5 through Class 1 actually mean, the accuracy range behind each, and how to put the framework to work on a real bid.

If you've spent any time in cost estimating, you've heard estimates sorted into "classes" — a Class 3 budget estimate, a Class 5 concept number. But the labels are often used loosely, and the actual framework behind them carries more useful information than most people take advantage of. This guide explains the AACE estimate classification system in plain terms, and shows how it connects directly to how much confidence your number deserves.

What the AACE classification system is

AACE International (the Association for the Advancement of Cost Engineering) publishes a standard framework that sorts capital project cost estimates into five classes, from Class 5 down to Class 1. The system replaced older, vaguer terms like "order-of-magnitude" and "definitive" with a consistent scale tied to one thing: how well the project is actually defined.

The generic standard is AACE Recommended Practice 17R-97, with industry-specific supplements — most notably 18R-97 for the process industries. The core idea is the same across all of them: the more design definition you have, the more accurate your estimate can be, and the narrower its expected range.

The five classes

Classes are defined primarily by level of project definition — expressed as the percentage of full definition complete — which in turn drives the estimate's expected accuracy range and its typical purpose.

Ranges shown per AACE RP 18R-97 (process industries); other industries differ \u2014 see below. Actual ranges should always be confirmed by risk analysis of the specific project, not simply assumed from the table.

Why the ranges get tighter as class improves

The logic is intuitive once you see it. A Class 5 concept estimate is built on almost nothing — a rough scope, a few benchmarks, a lot of judgment. The real cost could land almost anywhere, so the honest range is enormous. A Class 2 estimate, by contrast, sits on substantial engineering, quantified takeoffs, and firm pricing. There's far less room for the final number to surprise you, so the range collapses inward.

This is why quoting a tight range on a poorly-defined estimate isn't confidence — it's an error. If you're at Class 5 definition but reporting Class 2 accuracy, you're not being precise; you're hiding risk that's genuinely there.

The bands vary by industry

Here's a nuance most quick references skip: AACE doesn't publish one universal table. The generic framework (17R-97) is adapted into industry-specific recommended practices, and the accuracy bands differ meaningfully between them. The class definitions stay the same \u2014 Class 5 is always a rough concept, Class 1 always near-complete \u2014 but how wide the expected range is depends on the kind of project.

The pattern is intuitive once you see it. A commercial building is a known quantity \u2014 thousands get built, the cost drivers are familiar, so even a concept estimate has a relatively contained range. A first-of-a-kind process plant or a nuclear facility carries enormous scope uncertainty, and its bands reflect that: a nuclear Class 5 concept can run +200% over. Using the wrong industry's table can badly misstate your risk \u2014 a building estimator applying process-industry bands would carry far too much spread, and a process estimator using building bands would understate theirs.

What about Class 1?

Class 1 estimates are the most defined \u2014 built on near-complete design with known quantities. In practice, many estimators treat Class 1 as a check estimate or closeout validation, done to confirm pricing against actuals rather than to forecast an uncertain future. When quantities are essentially known, there's little left to model probabilistically; you're validating, not forecasting.

Connecting AACE class to a real risk analysis

Here's where the classification stops being a filing label and becomes a working tool. Start from the right industry's table, match your estimate's class to the appropriate band, decide where within that band your project sits \u2014 tighter if scope is clean and low-risk, wider if it's complex or poorly defined \u2014 apply it across your line items, then refine individual trades based on what you specifically know.

From there, a Monte Carlo simulation turns those ranges into an actual cost distribution — P10, P50, P90 — and a defensible contingency figure. The AACE class sets the starting spread; the simulation earns the final number. That pairing is exactly how a modern estimate moves from "what class is this?" to "what should we carry, and why?"