The CSI MasterFormat guide for general contractors.

What is CSI MasterFormat, practically?

MasterFormat is a classification system maintained by the Construction Specifications Institute (CSI) and Construction Specifications Canada (CSC). It organizes construction information into 50 top-level divisions — Division 01 (General Requirements) through Division 49 (Transportation Design and Construction) plus Division 50 (Site and Infrastructure). Each division contains sections identified by a six-digit code.

On bid day, MasterFormat is what lets you answer a deceptively simple question: are these two bids for the same scope? Without a shared taxonomy, “electrical” from one sub might include fire alarm while another sub's “electrical” specifically excludes it. With MasterFormat, Division 26 is Electrical, Division 27 is Communications, Division 28 is Electronic Safety and Security — and the ambiguity goes away.

The divisions that matter most on commercial GC bid days

In practice, a vertical commercial GC deals with a concentrated set of divisions on most projects. These are the ones that drive the majority of bid volume:

Percentages are rough planning benchmarks for mid-rise commercial and should not substitute for your own historical data.

Divisions vs. subdivisions vs. sections

MasterFormat uses a three-level hierarchy that trips up a lot of estimators because the CSI documentation uses the terms interchangeably with the actual numbering. The clean way to think about it:

• DivisionThe top level: 2-digit code. Example: 09 Finishes.
• SubdivisionA grouping within a division, identified by the first 4 digits of a section code. Example: 09 20 Plaster and Gypsum Board.
• SectionThe specific spec, identified by the full 6-digit code. Example: 09 29 00 Gypsum Board.

For bid-day leveling, subdivisions are usually the right level of granularity. Section-level (6-digit) classification is more than most subs provide and rarely worth re-classifying by hand. Division-level is too coarse — Division 09 Finishes lumps drywall, acoustical ceilings, and flooring together, which loses the comparison signal.

MasterFormat vs. UniFormat: when to use which

UniFormat organizes by building element (foundations, shell, interiors, services) and is the right tool for conceptual estimating at schematic design when you don't know what's behind the walls yet. MasterFormat organizes by work result (the actual installed system) and is the right tool once bids start coming back and you need to compare scope.

Most GCs use UniFormat for conceptual estimating and MasterFormat for bid leveling. They're not competing standards — they're complementary, and mature estimating teams crosswalk between them as a project moves from schematic design to bid.

The classification problem: why most GCs get it wrong

Subs don't send you MasterFormat-classified bids. They send you a PDF with line items written in their own language — “framing,” “rough carpentry,” “interior partitions.” One sub's “framing” might be Division 05 (steel framing), another's might be Division 06 (wood framing), a third's might span both.

The classification problem is what turns a 30-minute bid comparison into a 3-hour re-keying exercise. You can't build useful benchmarks on raw line-item descriptions because every sub describes the same work differently. Classification is the foundation of everything downstream — comparison, benchmarking, scope gap detection.

Three approaches:

1. Make the sub do it. Require MasterFormat-coded bids in your ITB. Works if you're a large GC with market power. Doesn't work if you're one of 20 GCs the sub is bidding.
2. Have an estimator do it. Manual re-classification while leveling. Accurate but expensive — this is the 3 hours per bid package most GCs eat.
3. Let AI do the first pass. An AI trained on MasterFormat can classify line items with high accuracy for standard trades, and flag low-confidence items for human review. This is where PreconIntel's Bid Inbox fits.

Using MasterFormat for cost benchmarking

Once your historical bid data is classified, the benchmarks practically build themselves. The useful unit to benchmark is almost always at the subdivision level — Division 23 HVAC per square foot is too noisy across project types, but 23 07 Thermal Insulation per linear foot of ductwork is a number you can trust across similar projects.

Benchmarks compound. Every parsed and classified bid adds to your history. After a year of bid activity a GC typically has thousands of benchmarked data points across their top 10 divisions — and that's the moat your competitors don't have unless they're also classifying systematically.

Common mistakes we see GCs make

• Classifying too deep. Section-level (6-digit) classification is more detail than you need for bid leveling and more than most sub bids support. Subdivision-level (4-digit) is the sweet spot.
• Using old MasterFormat. MasterFormat was restructured in 2004 and has had updates since. If your division library still treats Division 15 as Mechanical and Division 16 as Electrical, you're on pre-2004 MasterFormat and your benchmarks won't crosswalk with current bids.
• Treating MasterFormat as sacred. It's a classification system, not scripture. If your work includes a lot of solar PV and you want to carve it out of Division 26, do it — just document the rule and apply it consistently.
• Not classifying at all. By far the most common mistake. “We'll add it later” becomes “we have ten years of bid data and no benchmarks.”