The first question on nearly every RFQ that lands on our desk is which machine the part goes on. The answer usually looks obvious: it is round, so it goes on a lathe; it is boxy, so it goes on a mill. That rule is right about eight times out of ten and quietly costs money the other two. A shaft with a wrench flat, a flange with a bolt circle, a housing that is mostly turned but needs a milled pocket on the face — each of these crosses the line between the two processes, and the expensive mistake is usually forcing a part onto the wrong machine because it "looks" like a turning job or a milling job. This article is the working version of how to read a print and pick the process that costs less, holds tolerance, and does not leave you paying for machine time on features a different machine would cut nearly for free.
The geometry test comes first
Milling and turning remove metal in opposite ways, and that single difference resolves most parts in seconds. Turning spins the workpiece and feeds a stationary tool into it, so every cut produces a round, concentric surface by definition. Milling spins the tool and feeds the part through it, so it can reach any orientation and cut flat faces, pockets, slots, and holes anywhere on the part. One process generates surfaces that revolve around a single axis; the other generates everything else.
If the part is a body of revolution — a shaft, a bushing, a sleeve, a pin, a threaded stud — it is a turning job. The lathe generates the outside diameter in one pass, bores the inside diameter concentric to it, and faces both ends square to the axis without a second setup. That concentricity is the whole point. A turned OD and ID share the same spindle axis, so runout between them is limited by the machine and the tool, not by how well you re-clamped the part. A milled circle, by contrast, is interpolated by moving two axes at once, and its roundness depends on backlash, servo following error, and tool deflection — good enough for a bore that only locates a bolt, wrong for a bearing seat that runs at speed.
The test is simple: does the part's most important feature — the one with the tightest tolerance, the one that mates — revolve around a single axis? If it does, start with the lathe and add milling only for what is off-axis. If the critical features are flat, orthogonal, or spread across several faces at angles to each other, it is a mill job. The rest of the decision is how far you can push that first answer before it breaks.
What each process actually gives you
Turning's strength is concentricity and surface finish that come almost for free. Because the tool is held rigidly and the part rotates at high surface speed, a lathe will typically hold ±0.01 mm on a diameter without strain and land a 0.8 to 1.6 µm Ra finish on a turned OD in a single finish pass. A good CNC lathe bores the ID on the same axis it turns the OD, so you get concentricity of 0.01 to 0.02 mm between them as a matter of course — a number a mill has to fight for. Turning is also the cheapest way to peel a lot of metal off a round bar, because the tool stays engaged and the chip never stops.
Milling's strength is reach and versatility. A three-axis mill gets at five of the six faces of a cube in one or two setups, cuts pockets and bosses, drills and taps holes at any position in the plane, and holds flatness and parallelism on a face that a lathe can only touch by adding live tooling. Milling holds its own on tolerance — a good mill holds ±0.01 mm on a pocket and a few hundredths on feature position — but flat faces and orthogonal relationships are its home turf, not concentric diameters.
| Aspect | CNC Turning | CNC Milling |
|---|---|---|
| Geometry | Bodies of revolution: shafts, bushings, pins | Prismatic parts: plates, housings, brackets |
| Concentricity | OD and ID share one axis — 0.01–0.02 mm typical | Interpolated bores — depends on servo and tool |
| Surface finish | 0.8–1.6 µm Ra, often free | 1.6–3.2 µm Ra, stepover-limited |
| Typical tolerance | ±0.01 mm on diameter | ±0.01 mm on feature position |
| Best for | Long runs of round parts, bar-fed | Complex faces, pockets, holes at any angle |
| Metal removal | Constant engagement, high volume | High for aluminum with HEM toolpaths |
Neither process is "better" in a vacuum. The table matters only when you read it against a specific print, because a tolerance or a feature in one row will override everything else in the other five. A part with a bearing bore and a flat mounting face needs both, and the real decision is how to sequence them.
The hybrid parts that break the simple rule
The interesting parts are the ones that do not fit cleanly into either column. A shaft with a milled flat for a set screw is still a turning job, but it needs a second operation on a mill — or a lathe with live tooling. A flange is turned, but its bolt circle has to be drilled, and the question is whether to drill it with the lathe's live tools or in a mill fixture. A housing that is 80 percent turned and 20 percent milled is where shops genuinely argue, and the argument is usually decided by how tight the relationship is between the turned feature and the milled feature.
If the milled detail must be located relative to the turned axis to within a few hundredths — a keyway clocked to a hole pattern, a pocket on the face that has to sit concentric to a bore — doing it on the same machine is worth the cost. A mill-turn center with live tooling keeps the part on one axis and never re-clamps it, so the relationship between the turned and milled features is preserved by the machine instead of by a fixture you re-set between operations. If the milled detail is cosmetic or loosely located — a wrench flat, a logo pocket — a second operation on a mill is cheaper, because you are not paying mill-turn rates for a feature a three-axis mill cuts in minutes.
The rule of thumb we use: the tighter the relationship between an off-axis feature and the bore or OD, the more a mill-turn center or a lathe with live tooling pays for itself. When the tolerance between the two is loose, split the part across two cheaper machines and bank the difference.
Material and tolerance change the answer
Material pushes the decision in directions the print does not always show. Hard, stringy, or work-hardening materials favor turning wherever you can, because a lathe's single-point tool cuts at a constant chip load and does not shock the edge the way an interpolating mill does. Inconel and 316 stainless are both easier to turn than to mill, all else equal — the tool enters once, cuts steadily, and exits, instead of hammering in and out of the cut on every revolution. Soft, free-machining materials — 6061 aluminum, brass, acetal — mill beautifully, and for complex aluminum parts the speed of a mill with a high-efficiency toolpath often beats a lathe outright.
Tolerance does something similar. A bearing bore with a 0.005 mm tolerance wants to be turned or bored on the lathe, because boring on the same axis as the OD is the reliable way to hold position and roundness in one controlled cut. A flat face with a 0.01 mm flatness callout wants a mill with a face mill and a good finish pass, not a lathe facing operation that leaves feed marks in a spiral pattern. The rule is to match the tightest tolerance on the print to the process that generates that surface in a single, controlled cut, and let the rest of the part follow.
Volume, setup, and cost per part
For one-offs and small lots, the decision is driven by setup count. Fewer setups means lower labor and less chance of tolerance stack-up from re-clamping, so a part that can be done in one lathe setup wins over a part that needs three mill setups. For volume, the picture flips: a lathe can run bar-fed and lights-out, feeding raw bar and dropping finished parts for pennies of labor each, while a mill tends to need per-part fixture cycles and some operator attention. A 10,000-piece pin is a lathe job no matter what the print looks like; a 10,000-piece aluminum housing with pockets is a mill job on palletized fixtures, no matter how round one corner of it happens to be.
| Factor | Turning wins when | Milling wins when |
|---|---|---|
| Setup count | Round part done in one chucking | Five of six faces in one or two setups |
| Labor | Bar-fed, lights-out, low touch time | Fixture cycles, more operator time |
| Volume | 1,000+ pins, bushings, studs | Complex housings on pallets |
| Cost driver | Material size vs. finished part | Setup count and cycle time |
| Typical loser | Parts with off-axis details only | Round parts forced onto the mill |
The cost tables do not replace quoting; they narrow it down. When a part genuinely straddles the line, we quote both ways and let the price make the argument, because the same part machined on a lathe and on a mill can land 30 to 40 percent apart even though both meet the print.
A selection checklist that runs before the quote
Every line below is on this list because a specific job taught us it was necessary. It is the same list we run before we pick a machine for any new part.
- Find the critical feature first. The mating surface with the tightest tolerance decides the process; everything else follows it.
- Does it revolve around one axis? A shaft, bushing, or pin with a tight bore is a turning job unless something off-axis overrides it.
- Is concentricity the callout? Turned OD and ID share a spindle axis; a milled bore has to be interpolated. Match the surface to the process.
- Check the material. Work-hardening alloys like Inconel and 316 turn easier than they mill; aluminum and brass mill beautifully.
- Weigh the off-axis details. Tight relationship to the bore or OD means live tooling or a mill-turn center; loose details mean a cheap second op.
- Count the setups. Fewer setups beats more setups on a one-off; bar-fed beats fixtures on a big run.
- Match volume to the machine. A lathe runs lights-out on bar stock; a mill runs pallets. Pick the one built for your quantity.
- Quote both ways when it straddles. The same print can land 30–40 percent apart between a lathe and a mill. Let the price decide.
- Do not force a round part onto the mill. Interpolating a circle costs more and holds roundness worse than turning it.
- Do not force a prismatic part onto the lathe. If it needs pockets and angled faces, the lathe is the wrong tool no matter the volume.
"A customer once insisted we mill a 316 stainless bushing because they thought the lathe was only for simple parts. It took three setups and the concentricity kept drifting out of spec. We re-quoted it on the lathe, turned and bored it in one chucking, and the part came out better and cost about 40 percent less. The print never changed — only which machine it went on."
Milling and turning are not competing; they are complementary, and the skill is knowing which one owns the critical feature. Read the geometry first, then the material, then the tolerance, then the volume — in that order — and the process almost chooses itself. When it does not, quote both and let the number settle it. The wrong machine does not just cost more; it quietly erodes the tolerance you promised, and no amount of inspection fixes a process that was chosen badly on day one.
The short version of this decision lives on our CNC milling and CNC turning pages, with the machine fleets and tolerances behind each.