Single-Stream Recycling Runs 27% Contamination, and That's Where the Economics Break

The disc screens on a single-stream line tell you what the collection brochure won't. Film wraps the shafts, glass fines grind into everything, and a load that left the curb looking like recycling lands on the tip floor as maybe three-quarters recycling and one-quarter garbage. Nationally, about 17% of what goes into the residential recycling cart is contamination, per The Recycling Partnership's 2024 accounting, and single-stream programs sit well above that, into the mid-to-high 20s. Single-stream recycling made participation easy by making the sort hard, and the cost of that trade doesn't land where the savings do.

I've spent 20 years on the thermal side, taking in what MRFs can't sell. So my read on single-stream vs dual-stream comes from downstream, where contamination finally turns into somebody's disposal bill. The savings from going single-stream sit on the collection side: one truck, one cart, higher setout rates because residents don't have to think about it. The costs sit at the MRF and past it, in sort labor, downgraded bales, and residue that rolls to a landfill or a boiler. None of it shows on the collection invoice. And that's how these programs get chosen wrong.

The convenience is a cost transfer, not a cost cut

Single-stream won North America because it moves the number cities actually watch: setout rates and tons collected. Put everything in one cart and people recycle more. But "more" includes more of what shouldn't be there. That's wishcycling, the hopeful garden hose or the greasy pizza box that goes in the blue cart because it feels like it belongs. The EPA's last full accounting put U.S. municipal waste at 292.4 million tons in 2018, roughly half of it landfilled. Every point of contamination you take in at the curb is a point of landfill diversion you don't actually bank, because it leaves the MRF as residue you pay to bury anyway.

It's the same mistake I keep flagging in these pages, most recently in a piece on the beliefs about waste conversion that quietly cost operators money: the assumption that a dirty input is a downstream problem you can sort your way out of. You can, partly. You just pay for it every ton, forever.

Contamination rate: the number nobody measures the same way

Operators fool themselves on the recycling contamination rate itself. What you report depends entirely on where you pull the sample and how. Grab it off the inbound tip floor, you get one number; pull it after the pre-sort, you get a flattering one; weigh the residue at the back end and you get the real number, which is usually worse. I hold the same view here that I hold about most plant "efficiency" problems, that what you're really measuring is instrumentation and sampling drift as much as the process. The residue pile tells you what the inbound sample gate won't.

Actually, that statewide contamination figure is dragged up by a handful of bad programs; a single-stream MRF with real route-level education and a decent front end can hold contamination in the mid-teens. The material makes the difference. Glass is the villain: once a wine bottle shatters on the tip floor, the fines migrate into the fiber and the colored plastic, and contamination in those specific streams can run up toward 45%, by the University of Florida's numbers. Dual-stream keeps glass and fiber apart at the curb, so that particular disaster never starts.

Cost, sorted by where it lands

The at-a-glance version first, then the parts that matter.

The contamination spread there is the University of Florida's, drawn from Florida MRFs between 2007 and 2019; the processing-cost line comes from a National Academies review, and the capital-cost gap from the Institute for Local Self-Reliance. Read it by where the money moves, not by which column wins.

Single-stream cuts what you spend at the curb and adds it back, with interest, at the sort line. The thing that turned contamination from an annoyance into a hard line item was China's National Sword policy, which in 2018 set an import contamination limit of 0.5%. Bales that used to clear stopped clearing, and revenue for contaminated grades like mixed plastic dropped by as much as 70%, per ILSR. Handling contamination now runs a MRF on the order of $140 a ton, between the extra sorting labor and the disposal of what you screen out, per industry survey data.

Glass earns its own bill. Washington, DC was paying $120 a ton to process recyclables in 2019, plus a $25-a-ton surcharge just for glass, which was a fifth of the stream by weight. That surcharge is the price of single-stream convenience stated plainly: you collected the glass mixed, so now you pay to deal with it mixed. A dual-stream program that keeps glass in its own bin skips the surcharge and sells cleaner fiber on top of it.

Throughput and the equipment that eats the contamination

Contamination isn't only a revenue problem. It's a throughput problem, and this is the part the collection-side spreadsheet never sees. Wet, filmy, glass-laden material re-rates every machine downstream of the tip floor. Plastic bags wrap the disc screens and star shafts until someone climbs in with a knife. Glass fines sandblast the belts and the optical sorters' viewing windows. Even a good optical unit like a Tomra Autosort loses accuracy when the reading surface is dusted with fines, and a Nihot air separator tuned for clean fiber (and they're all tuned for clean fiber on the brochure) chokes on wet contamination. Rate the line at nameplate and you'll be let down; in my experience the OEM throughput number runs 10 to 15% optimistic by default, and contamination is how you find out.

I learned that the expensive way. On the Baltimore RDF line in 2018, wet, film-heavy feed bridged the main feed chute and stalled the line at 91% availability for the better part of a month. The SCADA showed the line running. It wasn't. Not at rate. The fix wasn't a control tune or a new sensor; it was re-angling the hopper and hanging a bin vibrator so the material stopped arching over the throat. Contaminated wet material behaves like that on any screen or chute, MRF or RDF line, and no dashboard warns you before it does.

The better facilities are throwing vision at this, retrofitting AI-driven optical sorters and waste intelligence software that flags contamination in real time. It helps. But it's cleanup after a collection decision that already made the mess, and a robot pulling a shredded bag off a belt costs more than a resident not putting it in the cart.

Which one fits your program

So is dual-stream worth reviving? For some programs, yes, and the deciding factor isn't ideology, it's what your end market punishes. If your recovered fiber sells into mills that hold you to tight ISRI grade specs, dual-stream's cleaner bales are worth the second truck, because a load that fails grade gets paid as trash. If glass is a big share of your stream and you've got a local glass market, keep it separate; single-stream glass is mostly a disposal cost wearing a recycling label. But if your program's real problem is that people won't participate at all, single-stream's convenience buys tonnage you'd never otherwise capture, and you manage contamination with route education and front-end investment instead of by splitting the cart.

None of this holds cleanly at small scale, and the limits are worth naming. Not every program can cost-justify the front-end sorting or the optical retrofit that makes single-stream tolerable, and a town running a few thousand tons a year may not have two trucks to run dual-stream either; those programs live or die on contracted MRF terms, not on their own equipment. Data quality is the other soft spot. If your contamination rate is a number a hauler self-reports off the inbound scale rather than an audited residue weight from the back end, you don't really know which system you're running, and you're defending a figure that lies.

Whatever you choose, the contamination has to go somewhere. What the MRF screens out doesn't disappear; a share of it becomes fuel for waste-to-energy technology, which is where I usually meet it, and the rest gets buried. If your actual target is zero-waste-to-landfill, contamination is the enemy of that target, not a rounding error, because every contaminated ton is a ton of landfill diversion you claimed and didn't get.

Single-stream isn't a mistake. It's a bet that cheaper collection and higher participation beat cleaner bales, and for plenty of programs the bet pays. The trouble is that the people making the call see the collection savings on their own budget and never see the contamination cost, because it lands three trucks downstream on somebody else's ledger. That's why single-stream keeps winning the vote and losing the math. Weigh your own residue tonnage before you defend either one.

Sources & Notes

• The 27% single-stream, 18% dual-stream, and up-to-45% glass-and-fiber contamination figures come from the University of Florida's multi-year study of Florida MRFs (2007-2019).
• The national ~17% residential contamination figure sits inside The Recycling Partnership's 2024 State of Recycling report, which also pegs residential capture at 21%.
• Capital-cost gap, the DC glass surcharge, and the 70% revenue drop on contaminated plastic all trace to the Institute for Local Self-Reliance's dual- versus single-stream breakdown.
• Processing costs of roughly $112 and $60 a ton are from a National Academies analysis of U.S. recycling; the ~$140-a-ton contamination handling figure is drawn from Waste360's reporting on single-stream contamination.
• National generation and landfill share (292.4 million tons, 2018) are EPA Facts and Figures. The screens, chutes, and throughput margins are my own, from 20 years on RDF and thermal lines.