The Most Common Cause: Curves That Are Too Tight for Your Equipment
If your model train runs fine on straight track but jumps the rails every time it hits a curve, the culprit is almost always the curve itself, not a faulty locomotive. Every scale has a minimum curve radius that manufacturers design rolling stock around, and it's easy to build a layout tighter than that number without realizing it until trains start derailing.
For HO scale, the often-cited "minimum" is 18 inches, but that's really a bare survival number for short, four-axle equipment. In practice, most hobbyists find 22-24 inches noticeably more reliable, and anything running longer steam locomotives, passenger cars, or 50-foot-plus freight cars does much better on 27-30 inch curves or larger. Other scales scale proportionally: N scale layouts typically want at least 11-inch curves for reliable running, while O scale often needs 36 inches or more for anything beyond a small switcher.
If you're not sure what radius your curves actually are, measure from the center of the track to the center of the curve, not to the outer rail. A curve that "looks" gentle on a small layout can still be tighter than your longest car can comfortably handle, especially at the transition where straight track meets the curve.
Second Most Common Cause: Not Enough Weight on the Cars
Weight is what keeps a car's wheels pressed down onto the rail through a curve. A car that's too light can rock, bounce, or have its wheels momentarily lift just enough for the flange to climb over the railhead — and once that happens, it's off the track. This shows up most on curves because that's where the sideways forces on a car are highest.
The NMRA has a long-standing weight recommendation: roughly one ounce of weight for the car itself, plus half an ounce for every inch of car length. A 6-inch HO boxcar, for example, should weigh in the neighborhood of 4 ounces. If your cars are noticeably lighter than that, add weight low in the car — press-in stick-on weights or epoxied lead/steel work well — and keep it centered so you don't throw off the balance.
Empty flatcars, gondolas, and hoppers are the worst offenders here because there's nothing inside them to add natural weight. These are usually the first cars to derail on a curve, especially if they're coupled right behind the locomotive where pulling forces are strongest.
Third Cause: Track Alignment and Gauge Problems
Even a properly sized curve will derail trains if the track itself isn't laid correctly. A few specific problems to check, in order of how common they are:
- Kinks at rail joints. When you bend flex track into a curve, the rail ends at each joint can shift slightly out of alignment, creating a tiny kink. Run your finger along the inside of the rail through every curve — you're feeling for any point where it's not perfectly smooth.
- Tight or uneven gauge. Use a track gauge tool to spot-check width through curves, especially hand-laid track or older flex track that's been bent and rebent. Gauge that's even slightly narrow will pinch wheel flanges on curves before it causes problems anywhere else.
- Uneven roadbed. A dip or high spot under a curve twists the track just enough to lift one wheel. This is very common where a layout crosses a table seam or where cork roadbed wasn't glued down flat.
- Reverse curves with no straight section between them. An "S-curve" where one curve immediately reverses into another, with no straight track in between, puts extreme stress on long cars and couplers. Adding even 3-6 inches of straight track between the two curves fixes most S-curve derailments.
A simple diagnostic: run a single, well-weighted car by itself (uncoupled) slowly through the problem curve. If it derails alone, the issue is the track. If it only derails when coupled to other cars, the issue is more likely weight, coupler height, or a mismatched consist.
Fourth Cause: Coupler Height and Mismatched Equipment
Couplers that don't line up vertically between two cars will bind and shove sideways as they go through a curve, which is often enough to pop one car off the rails. This is especially common when mixing older and newer rolling stock, or different manufacturers, since coupler height standards have drifted over the years.
An inexpensive coupler height gauge (widely available for HO, N, and other scales) lets you check every car in under a minute each. If a car sits too high or too low compared to the standard, shim the truck or replace the coupler rather than leaving it — a single mismatched car in a long consist can cause derailments several cars away from the actual problem.
Also check that trucks (the wheel assemblies) swivel freely. A truck that's screwed down too tightly can't pivot into the curve properly, which fights against the track geometry no matter how well the curve itself is built.
How to Diagnose Your Specific Derailment Systematically
Rather than guessing, work through these steps in order — they're arranged from most likely to least likely cause:
- Note the exact spot where derailment happens every time. A derailment that happens at the same physical point on the track is almost always a track problem (kink, gauge, uneven roadbed). A derailment that happens randomly at different points is more often a weight or coupler problem.
- Test with a single car, then with your full normal consist, to isolate whether it's a track issue or an equipment/weight issue.
- Check curve radius against your longest piece of rolling stock, not your average car.
- Inspect and clean the wheels and track — built-up dirt or oxidation on either can reduce the contact and traction that's helping keep the car on the rails through a curve.
- Weigh your lightest cars and compare against the NMRA guideline above.
Most layouts that seem to have a "cursed" curve turn out to have two smaller issues stacking on top of each other — say, a slightly tight radius combined with one underweight car — rather than one dramatic single cause. Fixing just one of the two often isn't enough, which is why systematic testing beats trial and error.
Getting the Track Plan Right From the Start
All of the fixes above are worth knowing, but the easier path is avoiding tight, awkward curves and mismatched track transitions in the first place — which mostly comes down to planning the layout properly before any track gets nailed or glued down. This is where a lot of beginners struggle, since it's not obvious how much radius a given locomotive actually needs, or how to lay out curves, transitions, and grades so equipment runs reliably instead of fighting the track from day one.
If you're setting up a layout for the first time, or troubleshooting one that's never run quite right, a structured beginner's course on model railroading fundamentals — covering track planning, benchwork, wiring, and scenery — can save a lot of the trial-and-error most people go through learning this hobby by feel. It's worth a look if derailments have you second-guessing your layout design rather than just chasing one bad curve.