How to Get Better First Layer Adhesion: The Settings and Checks That Actually Work

If your prints lift, curl or simply refuse to stick, the fix is almost always in the first 0.25 mm of plastic. Get that bottom layer right and the rest of the print tends to look after itself. Get it wrong and you’re scraping spaghetti off the bed an hour later. This guide covers how to get better first layer adhesion the practical way: the bed checks, the slicer values and the material-specific temperatures that genuinely move the needle — no hype, just the numbers that work.

We’ll go in the order that actually matters: get the mechanics right first, then tune the slicer, then dial in temperatures and cooling. Skipping straight to glue sticks before you’ve checked your Z-offset is how you end up chasing the same problem forever. For the slicer steps below we use our own ATN Slicer — a free OrcaSlicer-based slicer with our AI print-doctor built in, so it flags bad first-layer settings the moment you slice. The setting names and locations are identical to OrcaSlicer, so PrusaSlicer and OrcaSlicer users can follow along too.

Bed levelling and Z-offset are not the same thing

This trips up a huge number of makers, so let’s be precise. Bed levelling makes the print bed parallel to the plane the nozzle moves in — the nozzle should sit at the same height above the bed across the whole printing area. Z-offset tells the printer exactly how far the nozzle is from the bed when Z reads “0.” You can have a perfectly level bed and still get terrible adhesion because your Z-offset is wrong, and vice versa.

Here’s what each failure looks like:

Nozzle too high: filament lands as round beads with barely any contact, doesn’t really stick, and can be knocked off easily.
Nozzle too low: the nozzle scrapes the bed and blocks normal extrusion — you’ll see thin, see-through lines or gaps in the first layer.

The paper test, done properly

For a manual bed, the paper test still works if you do it right:

Preheat the bed and nozzle to your typical operating temperatures. This accounts for thermal expansion — a cold bed levels differently to a hot one.
Move the print head to one corner. Slide a standard sheet of paper (~0.1 mm thick) between nozzle and bed.
Adjust that corner’s levelling screw until you feel slight resistance when moving the paper.
Repeat at all four corners, then do the whole thing again. Adjusting one corner shifts the others, so it usually takes two full rounds to get the bed truly flat.

If all four corners feel right but the centre is loose, your bed may be warped. Cheaper beds — and any bed put through frequent heating and cooling cycles — can warp over time. A flat glass bed is often the simplest cure.

Fine-tuning the squish with baby stepping

Once the bed is level, you tune the actual squish with live Z-offset adjustment, or “baby stepping.” It’s been in Marlin and Repetier for years but usually isn’t enabled by default. It lets you nudge the Z axis up or down mid-print to perfect the first layer.

Marlin: Tune → Babystep Z. Each click is typically 0.025 mm.
Klipper: use the Z-offset buttons in Fluidd or Mainsail. Each click is usually 0.005 mm or 0.010 mm.

On most consumer setups, a more negative offset means the nozzle is closer to the bed and you get more squish — but direction conventions genuinely differ between firmware and probe setups, so verify on your own machine before committing. When the first layer looks right, save it: Marlin uses Control → Store Settings, Klipper uses SAVE_CONFIG. If you don’t save, the offset resets on the next power cycle and you start over.

How do you know it’s right? The lines should be well-adhered and slightly flattened, with no gaps and no excessive “elephant’s foot.” Per Ellis’ Print Tuning Guide, you should still clearly see the individual lines — if the surface is completely smooth, you’ve squished too much.

Slicer settings that make the first layer stick

Mechanics sorted, this is where you buy yourself margin. A thicker, wider, slower first layer is far more forgiving of small imperfections. We’ll give the paths in the ATN Slicer — because it’s OrcaSlicer-based, every setting name and location below is the same in OrcaSlicer, and the equivalents in PrusaSlicer are noted where they differ. The bonus with the ATN Slicer is that its built-in pre-flight engine warns you if any of these first-layer values are off the moment you slice, before you waste filament.

First-layer height and width

Height: a thicker initial layer holds more heat, so it has more time to bond before cooling. In the ATN Slicer, set it under Quality → Layer height → Initial layer height. For a standard 0.4 mm nozzle, around 0.24 mm is a good forgiving starting point. Advice varies across the 0.2–0.28 mm range, so treat it as a tuning band rather than a magic number. For larger printers, Ellis recommends 0.25 mm or greater — thicker first layers are simply less sensitive.
Width: push first-layer extrusion width to about 120% of nozzle diameter — that’s 0.48 mm for a 0.4 mm nozzle. Set it under Quality → Line width → First layer. Wider lines press into the bed and fill gaps better.
Flow: a small bump to the first-layer extrusion multiplier (Quality → Precision area) makes the layer thicker and stickier.

First-layer speed

Slow down. In the ATN Slicer it’s under Speed → Initial layer speed (PrusaSlicer calls it First layer speed). A typical first-layer speed for PLA and ABS is around 20 mm/s. A slower first pass means the filament sticks where it’s deposited rather than being dragged by the moving nozzle. If you’re still warping at 20 mm/s, drop to 10–15 mm/s.

If you want a sanity check before you commit a long print, slice in the ATN Slicer and its pre-flight engine flags first-layer issues right beside the gcode preview — or use our standalone G-code checker for the same thing in the browser. Either way it flags first-layer issues before they cost you a reel of filament. For exact, slicer-ready numbers, our breakdown of the best first layer settings in PrusaSlicer is the fastest route to a sticky base. Tip: the ATN Slicer is a free Windows download — grab it and let the AI print-doctor catch the mistake before you press print.

Cooling: fans are the enemy of adhesion

You want the plastic molten for as long as possible on that first pass. As a rule, lower or fully disable the part-cooling fan for the first few layers to keep the base warm, then ramp it up afterwards. In the ATN Slicer (and OrcaSlicer) these live under the Cooling section of the filament settings; PrusaSlicer puts them under Filament Settings → Cooling.

PLA: fan off (or very low) for the first layer or two, then bring it up.
PETG: fan completely off for the first two layers, then ramp to roughly 30–50% for the rest. (Exact PETG figures are vendor-specific — 30–50% is one common recommendation.) If you’re battling wisps and strings on PETG, our PETG stringing fix for OrcaSlicer covers the cooling and retraction settings that actually work — and they apply identically in the ATN Slicer.
ABS/ASA: very little part cooling at all. Strong airflow changes the temperature too quickly and causes warping. An enclosure helps enormously here.

Bed temperature by material

Temperature is the other half of cooling. Too cold and nothing tacks down; too hot and you invite elephant’s foot. Good starting points:

PLA: around 60°C gives optimal surface tack — the first layer grips well but still releases cleanly when the plate cools. Some setups run happily at 50–60°C.
PETG: PETG can stick too well to certain surfaces and tear chunks out of the plate, while refusing to stick to others. If that sounds familiar, read our piece on PETG first-layer adhesion issues before you reach for the glue.
ABS/ASA: a hot bed plus an enclosure to keep ambient temperature stable and avoid drafts.

If warping rather than pure adhesion is your problem, our guide on how to fix warping in 3D prints has the exact settings — and a brim is often the quickest win, adding bed contact area to anchor the corners.

When you’re still stuck: diagnose it from a photo

If you’ve levelled, tuned the Z-offset, fattened the first layer and matched your temperatures and it’s still not sticking, stop guessing. The fastest route is to slice in the ATN Slicer and open the Diagnose and Ask AI panels right beside the gcode preview — or upload a photo to our standalone Diagnose tool. Either way the vision AI identifies the defect and returns concrete, slicer-specific setting changes — and PrusaSlicer/OrcaSlicer users can also download an importable .ini patch. It’s the difference between a generic chatbot telling you to “level your bed” and a tool that tells you exactly which value to change. You can also read how our troubleshooting AI approaches problems like this. If you’d rather work from a visual catalogue of symptoms, our 3D print defect identification tool helps you spot and fix FDM failures fast.

FAQ

What is the ideal first layer height for better adhesion?

For a 0.4 mm nozzle, around 0.24 mm is a forgiving starting point — thicker layers hold more heat and bond better. Anywhere in the 0.2–0.28 mm range is reasonable; larger printers benefit from 0.25 mm or more. For PETG, ~0.28 mm helps stop the nozzle dragging strings of material. In the ATN Slicer (and OrcaSlicer) you set this under Quality → Initial layer height.

Should the cooling fan be on for the first layer?

No. Keep the part-cooling fan off (or very low) for the first one or two layers so the plastic stays molten long enough to bond, then ramp it up. For PETG, off for the first two layers; for ABS/ASA, keep part cooling minimal throughout.

Why does my first layer stick at the corners but not the centre?

That’s a classic sign of a warped bed. If all four corners pass the paper test but the centre is loose, the bed isn’t flat. A flat glass plate is the usual fix, or use mesh bed levelling to compensate for the dip.

Does a more negative Z-offset mean better adhesion?

On most consumer setups, a more negative offset moves the nozzle closer to the bed and increases squish, which improves adhesion up to a point — too much causes elephant’s foot. But the direction convention varies by firmware and probe, so confirm on your own machine before committing.