Gravel Under a Concrete Slab

How deep should the gravel be under a concrete slab?

Four inches, with three conditions almost every summary drops.

IRC section R506.2.2 is the governing residential requirement, and it reads: "A 4-inch-thick (102 mm) base course consisting of clean graded sand, gravel, crushed stone, crushed concrete or crushed blast-furnace slag passing a 2-inch (51 mm) sieve shall be placed on the prepared subgrade where the slab is below grade."

• It applies where the slab is below grade. A slab poured at or above surrounding grade is not covered by that sentence.
• There is an exception that removes it entirely. The code continues: "A base course is not required where the concrete slab is installed on well-drained or sand-gravel mixture soils classified as Group I according to the United Soil Classification System." If you are on clean sand or a sand-gravel mix, the code does not ask for a base at all.
• The only gradation imposed is a 2 inch sieve. Crusher run, #57, crushed concrete and ordinary bank run all pass that test. The code does not care which, which is why the rest of this page exists.

Outside the code, the industry numbers line up on 4 inches as a floor rather than a target. PCA and ACPA’s EB204P on subgrades and subbases recommends "Specifying a minimum subbase thickness of 4 in. (100 mm) for unstabilized (granular) subbases." USDA NRCS Design Guide 11 says "Minimum sub-base (stone) under slab shall be 4 inches" and then scales it to 6 and 8 inches as axle loads rise.

One correction worth making, because it is repeated everywhere. ACI 302.1R, the Guide to Concrete Floor and Slab Construction, does not recommend a 4 inch base course. It gives no base thickness at all, and defers the soil-support question to ACI 360R, which derives the thickness from the required modulus of subgrade reaction rather than naming a number. The 4 inches you keep reading about is the IRC’s and PCA’s, not ACI’s. The 4 inch figure that does appear in ACI 302.1R is the thickness of a fill course placed over a vapor retarder, which is a different layer doing a different job.

Use the slab calculator for the concrete itself, and slab thickness for how deep the concrete above this base should be.

What stone to use, and the #57 problem

Ask anywhere online what to put under a slab and the answer comes back "#57 stone." It is worth understanding what that actually designates, and then why the people who pour floors for a living disagree.

What #57 actually is

#57 is a gradation, not a rock. It is a size number from ASTM D448, mirrored in AASHTO M43, and the stone itself can be limestone, granite, diabase or anything else the local quarry sells. FHWA’s own test material for its open-graded aggregate study was diabase from Sterling, Virginia. The designation only tells you what comes through which sieve:

ASTM D448 itself is behind a paywall, so the table above is taken from two independent government reproductions that agree exactly: Maryland MDE’s reproduction of AASHTO M43 and Georgia DOT section 800.

The bottom two rows are the whole story. At most 5 percent of #57 passes the No. 8 sieve, which means it has almost no fines. That is what makes it open-graded: it drains like a colander, it does not hold water, and it reaches most of its density as soon as it is dumped.

Why that is the argument against it, not for it

The case for #57 is drainage. The case against it is what fresh concrete does when it meets a layer full of open voids.

The American Society of Concrete Contractors publishes a position statement on exactly this, Position Statement #37, and it names the material: "ASTM No. 57 and No. 67 coarse aggregates are examples of such materials." Its objection is mechanical. "Because of the openness of the permeable subbase [base] structure, mortar works its way into the voids during placement as the concrete is vibrated and consolidated." Paste that disappears into the base is paste missing from the slab, which finishes thinner than specified, and the mortar that keys into the stone restrains the slab as it shrinks.

ASCC’s own conclusion: "ASCC contractors believe that a dense-graded base covered by a 10 mil vapor retarder (as recommended in ACI 302.1R-04) is the best arrangement for interior slabs." And where a specifier insists: "ASCC contractors will place slabs on an open-graded base course if required by specification, but only with the acknowledgment that the risk of cracked or thin slabs is entirely the responsibility of the specifier."

ACI points the same direction. ACI 302.1R asks for a base with real fines in it, a clean fine-graded material with roughly 10 to 30 percent passing the No. 100 sieve. NRMCA’s CIP 29 repeats the figure: "A clean fine-graded, preferably crushed, material with about 10 to 30 percent passing the No. 100 sieve and free of clay or organic material is generally recommended." #57 has 0 to 5 percent passing the No. 8, a sieve four sizes coarser. It is not a near miss, it is the other end of the catalogue.

PCA gets there independently through permeability, advising against "unstabilized open-graded subbases with a permeability coefficient more than about 350 ft/day" and preferring "Free-draining subbases (subbases with a permeability between 50 and 150 ft/day)". Open-graded clean stone is far above that band.

So what should you actually use?

• An interior floor, especially one taking tile, vinyl, wood or any moisture-sensitive covering: a dense-graded base. Crusher run graded from 1½ inch to dust, which NRMCA describes as working well, or crushed concrete, which is dense-graded by nature and which the IRC names by name.
• A garage slab, shed slab, patio or anything exterior over heavy clay: open-graded stone is defensible, because drainage under the slab is a real problem and no flooring is going on top. This is practical advice rather than a code or ACI position, and ASCC would still rather you did not.
• If you use open-graded stone anyway: ASCC gives the compromise. "If open-graded bases are specified, the surface of these bases should be choked off with finer material." A thin layer of fines over the top closes the voids the paste would otherwise run into.
• Do not use concrete sand. NRMCA is blunt about it: "Concrete sand should not be used as it is easily displaced during construction."

There is one place where code positively asks for clean open stone under a slab, and it is not the floor. IRC Appendix F, radon control, section AF103.2 requires a gas-permeable layer of "A uniform layer of clean aggregate, a minimum of 4 inches (102 mm) thick ... material that will pass through a 2-inch (51 mm) sieve and be retained by a 1/4-inch (6.4 mm) sieve." #57 fits that window exactly. If you are in a radon zone, the absence of fines is the entire point, and the trade-off above resolves the other way.

Compacting the base, and why #57 cannot be Proctor tested

Two materials, two completely different compaction regimes, and specifications that regularly conflate them.

A dense-graded base is compacted to a percentage of Proctor density. PCA’s EB204P calls for "a minimum density of 95 percent of AASHTO T99 (ASTM D698) for unstabilized subbases." ACI 360R names both tests and gives "Nominal targets for these materials are from 90 to 95 percent of the modified Proctor dens[ity]." Those are not the same requirement. Modified Proctor applies several times the compactive energy of standard, so 95 percent Modified is a materially harder target than 95 percent Standard. Any specification that says "95 percent compaction" without naming D698 or D1557 is ambiguous, and it is worth asking which before you hire a testing lab.

An open-graded stone like #57 is not Proctor tested at all. There is not enough fine material for a Proctor curve to mean anything. FHWA-HRT-15-034 states the field practice directly: "In the field, compaction control for OGAs is typically a method-based specification, such as compact to non-movement or no appreciable displacement and assess with visual inspection." Its density is bracketed instead by the index tests ASTM D4254 and D4253, which for No. 57 give a minimum of 95.4 lb/ft³ and a maximum of 108.7 lb/ft³.

So "#57 does not need compacting" is too strong. It does get compacted, by method: run the plate over it until it stops moving. What it cannot do is produce a percentage for an inspector.

ACI 302.1R-15 suggests the practical check is not a density test either, noting that "proof-rolling results are much more indicative of the soil-support system’s ability to withstand loading than from the results of in-place tests", with corrective work where rutting or pumping exceeds half an inch.

Spread proud and compact down. The depth in the calculator above is the finished compacted depth. A dense-graded base loses roughly a fifth of its loose volume when it is rolled, so lay it in lifts of 4 inches or less, compact each one, and expect to spread noticeably more than the number you measured. Crushed concrete by the ton covers the loose against compacted weights in detail.

Where the vapor retarder goes

This is the part of the job with the most stale advice attached to it, because the guidance reversed and the old version never stopped circulating.

The current position is that the slab goes directly on the vapor retarder. ACI 302.1R-15 section 5.2.3.2: "the greatest level of protection for floor coverings, coatings, or building environments is provided when the vapor retarder/barrier is placed in direct contact with the slab." Laps overlap 6 inches.

It used to say the opposite. ACI 302.1R-96 required a blotter: a layer of granular fill between the retarder and the slab, meant to let the concrete dry downward as well as up. Then reports of moisture failures in flooring started arriving on exactly those jobs. As Peter Craig put it in Concrete International, "reports began to surface about moisture-related flooring failures occurring on projects where the fill course / blotter layer had been used." The problem is obvious in hindsight: a granular layer sitting on a sealed membrane has nowhere to drain, so rain, saw-cut slurry and curing water collect in it and feed the slab from below for years. An ACI task group formed in 1998, and the guidance changed in the 2004 and 2006 editions.

NRMCA still allows the blotter in one case, which is worth knowing because it is not simply wrong: where the floor will take no moisture-sensitive covering, a blotter of "a minimum 4 inch layer of compactable, easy-to-trim material" is acceptable, and "The granular layer should be dry prior to concrete placement to function as a blotter." The key word is dry. If you cannot guarantee that, the membrane goes against the slab.

What the membrane itself has to be

Here is a specification detail almost universally got wrong online. ASTM E1745 sets the permeance limit at 0.1 perms, and it is the same 0.1 perms for Class A, Class B and Class C. The classes differ only in strength:

Class A is not less permeable than Class C. It is tougher, which matters because a membrane that tears under a wheelbarrow is not a membrane. Note also that E1745 specifies no thickness in mils at all; it is a performance standard. The 10 mil figure people quote comes from building code, not from ASTM.

What the code requires depends on which code your town adopted

The IRC requirement has moved twice in three cycles, which is why two contractors can both be right and disagree:

The 2021 wording is verbatim "A minimum 10-mil (0.010 inch; 0.254 mm) vapor retarder conforming to ASTM E1745 Class A requirements with joints lapped not less than 6 inches (152 mm)". The 2024 reversal is described by NAHB’s summary of the cycle’s changes as allowing 6 mil polyethylene again and reversing the previous cycle. Check which edition your jurisdiction has adopted before buying membrane, because "code requires 10 mil Class A" was true for one cycle only.

The exceptions are stable across all of them, and they matter to most people reading this: no vapor retarder is required for garages, utility buildings and other unheated accessory structures, for unheated storage rooms under 70 square feet, or for "Driveways, walks, patios and other flatwork not likely to be enclosed and heated at a later date." If you are pouring a patio or a detached garage slab, the IRC asks for nothing here.

If you are insulating under the slab as well, the order of the layers changes again: under-slab insulation covers where the foam sits relative to the membrane and the base.

Putting it together

1. Strip topsoil and organics, then proof-roll the subgrade and fix anything that ruts or pumps more than half an inch.
2. Place the base in lifts of 4 inches or less, compacting each one. A dense-graded base goes down roughly a fifth proud of the finished depth.
3. Fine-grade the top. ACI 302.1R-15 notes a thin layer of graded granular material is normally used "to better control concrete’s thickness and to minimize friction between the base material and slab."
4. Vapor retarder directly under the slab, laps 6 inches, in the thickness your adopted code edition requires. No sand on top of it unless the floor will never take a moisture-sensitive covering and you can keep the sand dry.
5. Pour. The slab calculator sizes the concrete, and how to pour a slab covers the rest.

Sources

• ASCC Position Statement #37, Open-Graded Base Courses for Slabs-on-Ground, for the objection to #57 and #67 under interior slabs and the choke-off compromise.
• ACI 302.1R-15, Guide to Concrete Floor and Slab Construction, chapter 5, for direct contact between the vapor retarder and the slab, the 0.1 perm limit and the 6 inch laps.
• NRMCA CIP 29, Vapor Retarders Under Slabs on Grade, for the 10 to 30 percent passing the No. 100 sieve, the blotter conditions and the warning against concrete sand.
• PCA and ACPA EB204P, Subgrades and Subbases for Concrete Pavements, for the 4 in. minimum subbase, 95 percent of AASHTO T99, and the permeability band.
• FHWA-HRT-15-034, Strength Characterization of Open-Graded Aggregates for Structural Backfills, for method-based compaction of open-graded stone and the No. 57 index densities.
• Maryland MDE reproduction of AASHTO M43 coarse aggregate sizes, for the #57 gradation, corroborated against Georgia DOT section 800.
• USDA NRCS Design Guide 11, Floors (Slabs-on-Ground), for the 4 inch minimum and the thickness increases by axle load.
• Peter A. Craig, "No Sand Please!", Concrete International, for the history of the blotter layer and why the recommendation was reversed.
• ASTM E1745-17, Standard Specification for Plastic Water Vapor Retarders Used in Contact with Soil or Granular Fill under Concrete Slabs, Table 1, for the class requirements.
• IRC sections R506.2.2 and R506.2.3, and IRC Appendix F section AF103.2 for the radon gas-permeable layer.