Here's a number that should annoy you: 0.004 inches. That's roughly the width of a human hair — and it's the difference between a casting that slides into your assembly line on the first attempt and one that triggers a 12-week rework loop.

Now here's a second number: $137 billion. That's the approximate value of the global casting industry's shipments in 2025, with investment casting at about $16 billion and die casting at roughly $68 billion, per industry trade estimates. Tens of thousands of buyers every quarter stand at the same fork: investment casting or die casting? And high-tolerance components are exactly where the decision gets uncomfortable.

A third number gets quoted wrong constantly — ±0.005 inch for investment casting, ±0.001 inch for die casting. Both numbers are genuinely useful as a starting point. But they're dangerously misleading if you don't understand where they come from, what they actually permit on a real part, and how much you'll overpay for a tolerance you don't need.

This guide breaks down the full picture — linear tolerances, geometric tolerances, surface finish, and the cost behavior that actually drives your decision. By the end, you'll have a 5-step framework to stop guessing on your next RFQ.

The two tolerance numbers everyone quotes wrong

±0.005 and ±0.001. If you've compared casting quotes before, you've seen these two numbers on spec sheets, boilerplate, and sales decks. They look like answers. They're not.

What those numbers actually describe is a process capability band — a typical spread of variation a commercial foundry can hold on a small nominal linear dimension. And that's the catch. The number only means something in context. A ±0.005 tolerance on a half-inch boss is one thing. On a twelve-inch manifold body, it's a completely different promise.

Here's why the rule-of-thumb versions get dangerous. The tolerance a foundry can hold depends on:

• Part size. Bigger parts mean more cumulative shrinkage and more thermal variation.
• Feature direction. Dimensions within one die half behave differently than dimensions crossing a parting line or a movable core.
• Material. Aluminum and zinc shrink differently than steel and stainless. Each alloy has its own contraction curve.
• Tooling condition. A fresh die vs a die at shot 180,000 will not hold the same numbers.

So the next time a supplier writes "±0.005" on a quote, the right response isn't "thanks." It's "on which dimension, across which features, and what's your measured CpK?"

Why the process itself sets the tolerance

You're not buying a process. You're buying a dimension that has to survive a process. So it pays to understand where the variation actually comes from.

Investment casting starts with a wax pattern — an exact replica of your part, except it's plastic and it bends. That pattern gets coated with ceramic slurry to build a shell. The wax is melted out. Metal is poured in. And only after the shell is broken away do you see what you actually got.

Every layer in that chain adds a variable. The wax shrinks as it cools in the die. The ceramic shell expands during preheat. The metal contracts as it solidifies. On small parts, that stack-up stays manageable. On larger parts, the accumulated drift gets serious.

Die casting works differently. Two hardened steel die halves clamp shut. Molten aluminum is injected at pressures up to 10,000-20,000 psi. The steel die barely moves shot to shot. That repeatability is the strongest selling point of the process — and it's real. But the die is still a mechanical system. It wears. It thermal-cycles. The parting line develops flash over time. Movable cores and slides create their own seams and draft requirements.

Here's the punchline: the question is never "which process is more precise?" It's "which process can hold this feature on this dimension at this volume?" That's a different question, and the answer is different for every part.

Linear tolerance, side by side

Here's what commercial tolerances look like in practice, based on typical ranges quoted across RFQ data on our platform in 2025-2026:

Read that table twice, and notice what it doesn't say. Die casting is not "tighter everywhere." It's tighter on small dimensions, held within the same die half, measured parallel to the parting line. When you start crossing a parting line, you add typical mismatch of ±0.002 to ±0.004 inches on top of the linear number.

The other thing the table hides is that investment casting tolerances can be dialed in. Because the wax tooling is adjustable, a foundry can move the center of the tolerance band to favor a critical dimension. Say you have a bore seat that really matters. A good investment caster will tell you straight: "We can hold +0.003/−0.002 on that feature, but not all 14 dimensions on the print." That honesty matters more than any single number.

Think of it as a dartboard. Tolerance is the size of the group. Die casting throws a tight group. Investment casting can aim the group wherever the tooling allows. Both are useful. They're just different games.

Geometric tolerances: what actually breaks in assembly

Linear tolerances get all the attention. Flatness, straightness, parallelism, and concentricity are the features that actually cause rework loops. And they're the ones buyers forget to put on the drawing.

Typical as-cast flatness gives you a sense of scale:

• Investment casting: roughly 0.005-0.010 in per 4 inches of surface.
• Die casting: roughly 0.002-0.005 in per 4 inches as-cast.
• Concentricity and true position on bosses or fins: both processes need machining if you're chasing anything tight.

"The linear callout was fine on the print," says Rory Day, a Georgia aftermarket distributor who spends north of $1.5M on castings a year. "The flange was bowed eight thou and the whole batch went back. Nobody quoted flatness."

Notice what Rory said: nobody quoted flatness. The foundry quoted linear tolerances, the geometry wasn't called out, and everyone discovered the gap at the receiving dock. That's the expensive way to learn GD&T.

Two rules to avoid that fight:

Put the GD&T on the print before quoting, not after tooling is cut. If flatness or parallelism matters, call it out with a tolerance. A foundry can only hold what it's asked to hold. Ask for a geometric tolerance that doesn't exist on the drawing, and the answer is always "that's extra tooling and extra per-part cost" — which is code for "we didn't budget for it."

Never spec a tolerance across a drafted surface. Castings need draft angles to release from the die or the wax tooling. A wall with 2 degrees of draft doesn't have a single stable dimension — it has one at the top and another at the bottom. Buyers who don't account for this end up rejecting parts that are perfectly good.

Surface finish: the tolerance you can feel

Tolerance isn't just about size. It's about the surface you get. And on this metric, die casting has a genuine, measurable edge.

The reason is mechanical. Die casting forces molten metal against a polished steel die under extreme pressure — the part literally copies the die's smooth surface. Investment casting, on the other hand, copies the texture of a ceramic shell. That shell is refractory and porous, and the metal picks up that texture like a fingerprint.

Where does this matter for your part?

• Sealing surfaces. If a face needs a fluid-tight seal, both processes need a machining pass anyway. The as-cast finish difference stops mattering once the cutter runs.
• Fluid flow. Rough internal passages create turbulence and pressure drop. For hydraulic manifolds and valve bodies, die casting's smoother internal finish is a genuine advantage.
• Cosmetic parts. Visible brackets, housings, and covers look better straight out of a die casting. Investment castings usually need blasting or coating to look presentable.
• Fatigue performance. Surface asperities are where cracks start under cyclic loading. If fatigue life is a design concern, a machining pass or shot peening does more for you than switching processes.

A note on cost: specifying a finer surface finish than your part functionally needs is one of the fastest ways to inflate per-part price. A 63 Ra callout on an investment casting when 125 Ra would do — that's extra shell coats, extra processing time, and a higher scrap rate. Ask yourself what the surface actually needs to do before you put the number on the print.

The cost of tolerance: paying for precision you don't need

Tolerance decisions don't happen in a vacuum. They happen inside a cost structure, and the two processes have very different ones.

Tooling: Investment casting tooling is a wax injection die — smaller, simpler, cheaper. For most parts, you're looking at roughly $5,000 to $20,000 depending on complexity. Die casting tooling is a hardened steel die with ejectors, cores, and slides. Budget $50,000 to $150,000 or more, and expect more if your part has moving core features.

Per-part cost: Because die casting runs fast and hot on automated machines, its per-part cost is lower at volume — typically in the $3-8 per kg range for aluminum. Investment casting's gravity-pour, shell-build process costs more per shot — more like $8-20 per kg depending on alloy.

So you have a classic crossover. Die casting starts with a heavy tooling hit and pays you back on every part. Investment casting starts cheap and charges you more per piece, forever.

Where's the crossover? In our data, it typically falls between 5,000 and 30,000 parts per year, depending on part size and complexity. A small, simple bracket crosses over early. A large, complex housing with internal cored passages may never make economic sense as a die casting.

Run both numbers before you pick a lane: tooling amortized over your expected three-year volume, plus per-part cost, plus secondary machining, plus a realistic scrap allowance. The process that looks cheaper on the quote sheet is often the one bleeding you on the back end.

And don't forget the tolerance variable. Tighter tolerances mean more machining, more inspection, and more scrap on both sides. Every ±0.001 you tighten beyond what the part functionally needs is a tax on every single piece you'll ever buy.

Alloy, volume, and part size: the constraint filters

Before you even get to tolerance, two hard constraints will eliminate one of the processes for most parts.

Material

Die casting is limited to alloys that behave under high-pressure injection: primarily aluminum, zinc, and magnesium, with some copper alloys. The moment your print says stainless steel, carbon steel, titanium, or a nickel superalloy, die casting is off the table. Investment casting handles all of those. If your part is steel, this decision is already made.

Part size

Die casting machines have physical limits — shot weight, tie-bar spacing, platen size. Most aluminum die castings stay under 20 pounds, and large ones get expensive fast. Investment casting can produce parts up to several hundred pounds, though the typical sweet spot is under 10 pounds. A 14-inch housing is die-castable, but the tooling and machine size will crush your budget. Investment casting stays reasonable.

Internal geometry

Complex internal passages, curved cores, and intricate undercuts are investment casting territory. Die casting needs movable slides to form undercuts, which raises tooling cost and adds draft constraints. If your part has internal galleries, investment casting is usually the cleaner path.

A 5-step framework to pick the right process for your part

Here's the framework we use internally when a buyer sends us a print and asks "which one?" Run it in order, and it'll eliminate most of the guesswork.

Step 1: Define the true functional tolerances

Walk through the drawing and mark which dimensions actually matter for the part to function — bearing seats, sealing faces, mounting holes. Everything else gets a commercial band. Most buyers discover that only four or five features on a 30-dimension drawing genuinely need tight control.

Step 2: Map the GD&T before you send the RFQ

Add the flatness, parallelism, concentricity, and true position callouts now. If you send a print with only linear tolerances, you will get quotes based on only linear tolerances — and then you get to renegotiate at the PPAP stage. That's the most expensive way to add requirements.

Step

Need Help with Investment Casting vs Die Casting Tolerance: 2026 Data?

Get matched with 3 verified foundries in 24 hours. No markup, transparent pricing.

Request Quotes →