Mystery derailment gremlin

I would suspect truck (lack of) equalization or the inability of one truck on the rolling

stock to be able to tip a small amount to follow uneven track.

If it happens at the same spot it's probably the track . if it happens at  any old place perhaps the gauge of the engine?

If it happens over a switch, check the frog.  My frog came loose and that was the issue (although it was a 3-rail set up).

I have an Atlas PS4750 Hopper that does that. Noticed that one of the couplers, while at the proper height, was skewed ever so slightly. Fixed that and the derailments were reduced, but not eliminated. Need to take the car apart to check to see if the center sill has a twist in it.

You might have something like the above going on. Doesn't take much sometimes.

I'd agree if it occurs not at the same location each time, then one of the trucks needs better equalization flexibility.

On the track you could take a look at cross level or rate of change of cross level.  On the car or loco a frame twist or same problem in the trucks.  Anything that would percent all the wheels from sitting on the track with about the same weight on each one.  Another  issue is the trucks need to swivel freely so the car is not trying to go straight in a curve or continuing to curve after the track has straightened out after a curve.

Ok, so 1) check equalization, 2) check for frame twist. 

I'd check for knuckle alignment if it were the cars, but we know why our cars derail...usually too light, and in a switch.

Makes sense...considering a lot of the motive power dates back to somewhere between The Jurassic Period and 1948... and a lot of scratchbuilt stuff.

sure beats head scratching when the rail isn't the problem. thanks guys. i'll let you know what we find.