Used vs new industrial robots: what you really save.

Start with the share of the cell that's actually the robot. Universal Robots' co-founder put the arm at 25 to 30 percent of total system cost for traditional robots, and that's roughly what we see. Our robot integration cost guide walks through a machine tending cell where the arm is 28 percent of a $162,000 cell.

Now cut that arm's price in half. You've saved 14 percent of the cell, not 50. And you've probably added a little back. An older controller may need a different PLC interface, the integrator may be less fluent in that generation, and someone has to inspect, re-grease and re-master the arm before it goes in. The math below uses round, illustrative numbers.

We're not against used robots. We've seen them make a lot of sense. But the "half price robot" pitch only describes one line on the quote, and that line usually isn't the one that decides whether the cell pays back.

The downtime hours are the soft number here, so be honest with yourself about them. A well-refurbished arm with a known history may not stop any more often than a new one in its first few years. A tired arm with no records may stop a lot more. What you can't count on is the new-robot warranty to absorb the first surprise, and the five-year math has to carry that risk somewhere.

Resale value cuts both ways. A new arm will be worth more in five years than a used one, which narrows the gap further toward new. We left resale out of the chart to keep it simple.

To put a real number on your downtime rate, start with the contribution margin of what the cell makes per hour and whether you can recover lost hours on overtime. Our robot maintenance cost guide has a worked downtime example, and the robot ROI calculator will run payback both ways.

The robot safety standards were rewritten in 2025. ISO published new editions of ISO 10218-1 and 10218-2 in January, and A3 published the U.S. adoption, ANSI/A3 R15.06-2025, that September. We cover what changed in ISO 10218 and R15.06 were rewritten. The short version for a used-robot buyer is that your new cell's risk assessment is done against those documents, not against whatever was current when the arm was built.

That's where controller generation bites. Modern controllers offer safety-rated speed and position limiting, safe zones and safe I/O, which let an integrator shrink guarding or let people work closer to the cell. If your used controller doesn't have those options, or can't run them, the risk assessment may push you to more fencing, bigger footprints or a different layout. That's integration money.

The used robot inspection checklist

• Jog each axis slowly through its full range, then at speed. Listen for grinding, clicking or a rough spot that repeats at the same angle.
• Check backlash on the main axes with a dial indicator (Fig. 04). Compare with the maker's figure or a known good arm.
• Pull a little grease from the reducer outlets. Dark grease with metallic glitter means wear. Ask when it was last changed.
• Look at every joint seal for oil seepage and wet paint underneath.
• Cut servo power with a weighted tool and measure the drop.
• Inspect the cables where they move, especially through J1 and along the upper arm. Look for cracked jackets and taped repairs.
• Read the alarm history and the hours. Overload and collision alarms tell you how the arm was treated.
• Print the software options list and confirm the ones your application needs are there.
• Test every E-stop, the enabling device and the safety inputs.

Maintenance that's probably due

Assume nothing on the schedule has been done. FANUC's mechanical unit manual for the M-20iD family is a good example of what to plan for. It calls for replacing the arm's encoder batteries every year regardless of operating time, the reducer grease and gearbox oil every 3 years or 11,520 hours, and the mechanical unit cable and controller batteries every 4 years or 15,360 hours. Those intervals assume 3,840 operating hours a year, and the manual says to shorten them in proportion for heavier use.

So a ten-year-old arm that ran two shifts may be several grease changes and at least one cable replacement behind. Budget for all of it up front, then treat anything the seller can document as a bonus. Our robot maintenance schedule and cost guide lays out the full interval chart.

"Refurbished" on a listing can mean anything from a full teardown to a wipe-down and fresh paint. Paint is the one thing we'd rather a seller didn't do, honestly, because it hides the seepage and collision marks you want to see. Ask for the work list, part numbers of what was replaced, and the run-test record.

A good refurbisher will tell you what they didn't do, too. If the reducers weren't opened, that's fine for most applications as long as backlash and noise checked out. It just needs to be said.

FANUC's own service statement is unusually strong. It says it provides lifetime maintenance for its products as long as customers use them, and that it stocks over 3.5 million pieces of 19,000 types of repair part, including parts for models no longer made. That's a big reason used FANUC arms hold their value and a big reason integrators are comfortable building with them.

Support isn't the same as cheap or fast, though. A part for an old model may be in stock or may need to be sourced, and a lot of the value of a used arm depends on the controller it comes with. The M-20iD manual we cite, for example, pairs that arm with the R-30iB and R-30iB Plus controllers. Older generations still run fine, but the software options, safety functions and interfaces available for them differ, so get the exact controller model on the listing.

The same logic applies to ABB, Yaskawa, KUKA and the rest. Ask the maker, in writing, what support that controller generation still gets and what the lead time is on the big parts. A used arm whose maker has moved its controller to limited support is a bargain only if you'll replace it before the parts get hard to find.

Cobots are a newer market, so most used cobots are recent. That helps on support. The catch is that cobot cells often lean on safety-rated functions to run with less guarding, so confirm the used unit's software version supports the functions your risk assessment needs. For a primer on brands, see industrial robot arms compared and our FANUC vs ABB head-to-head.

Good fits we've seen include a second palletizer for a slower line, a spare arm kept on the shelf for a fleet of identical cells, a training cell for maintenance staff, and simple machine tending on a machine that already has slack in its schedule. In each case the cell can stop for a day without much pain and the plant already knows the platform.

Poor fits are the opposite. A first robot cell, where you're learning everything at once, is a bad place to also learn an old controller. A cell that paces a line turns every extra stop into lost output. And a layout that depends on modern safety-rated functions is a bad match for a controller that might not have them.

If you're on the fence, there's a middle path. Buy the cell new with a mainstream arm, and keep a used arm of the same family as a spare. You get warranty and current safety features where they matter, and a cheap way to cut downtime if the worst happens.

And if the real issue is capital, not the robot, look at the alternatives before you settle for an older arm. Our guide to robot leasing and robots as a service compares lease, buy and subscription on the same cell.