Freight train derailment caught on live stream - Pennsylvania, USA

[najaB]

On the afternoon of August 11, 2026, Norfolk Southern manifest freight train 33A, heading westbound, derailed on Horseshoe Curve near Altoona, Pennsylvania. This was caught live on the Virtual Railfan YouTube stream:

It sounds like the train goes into emergency braking around the 5 minute mark and the first sounds of derailment are around 5:30.

As an aside, are railways in the USA unsafe, or are UK railways unusually safe as we don't seem to have anything close to the number of accidents and incidents as they do, even accounting for relative size of the networks?

 

[ainsworth74]

As an aside, are railways in the USA unsafe, or are UK railways particularly safe as we don't seem to have anything close to the number of accidents and incidents as they do, even accounting for relative size of the networks.

Little of column A, little of column B.

US railroads are not as safe as the UK nor many other networks and have plenty of accidents and derailments (though most are low speed in yards, so not as dramatic as this). Obviously they have the inherent safety of railways which means they are still a safe form of transport in the US but not as safe as they should be. You can play with this site for all sorts of data (make sure to scroll all the way down to check some of the footnotes for context of some of the stats). But for instance in 2025 they had 983 derailments of which 296 were on the mainline so several derailments per week on the mainline.

 

[eldomtom2]

The safety culture on US freight railroads is extremely poor and practically Victorian in places. From what I've seen in discussions online every freight rail worker there has multiple tales about being pressured to ignore rules to get the job done quicker. There's absolutely no belief that the freight rail companies would ever place safety over profit.

 

[Gag Halfrunt]

IIRC tracks are generally maintained to the minimum standard required for slow freight trains. If a section of track is especially bad the usual tactic is to impose a "slow order" (speed restriction) if possible to avoid the need to repair it.

 

[Sir Felix Pole]

There was the apalling disaster at East Palestine, Ohio on the former Penn. R.R. in 2023. A tanker train of vinyl chloride and other hazardous materials derailed, causing a major confllagration. N & S seemed to be more concerned with getting the line open again than the long-term health of the residents, using some very questionable methods to clear the wteckage.

 

[Thames99]

How would the driver be aware that there had been a derailment dozens of wagons back on the train? Was he responding to the train 'feeling' heavier than it was or is there some kind of alarm or video system to keep an eye on the train?

 

[ainsworth74]

On the plus side they're so experienced at clearing derailments the train is already gone and the line appears to be re-opened. You can go the live stream:

Horseshoe Curve – Altoona, Pennsylvania, USA | LIVE Train Camera (PTZ)

ABOUT THIS LIVESTREAMVirtual Railfan’s camera at Horseshoe Curve, hosted by the Altoona Railroaders Memorial Museum, streams one of the most famous railroadi...

www.youtube.com

And scrub back through to watch them re-railing the wagons overnight. Looks like they were re-railed and on the move around 0940ish (EDT) today which considering the derailment was around 1520ish (EDT) yesterday is quite good going! They've had some on track plant going over the nearer track (next to the track the derailment was on) so I wonder if that is still shut. Had some tampers out there so I guess the impact next to that track of the derailed wagons disturbed the track and ballast.

How would the driver be aware that there had been a derailment dozens of wagons back on the train? Was he responding to the train 'feeling' heavier than it was or is there some kind of alarm or video system to keep an eye on the train?

Well it's Horseshoe Curve so may well have observed it directly as it's a big curve! I noticed that someone in orange is heading towards the cab at the rear the pair of locos so they also might have spotted it themselves and radioed the front of the train. Alternatively they might just have lost the brake pipe and that triggered the emergency brake automatically.

 

[stuving]

How would the driver be aware that there had been a derailment dozens of wagons back on the train? Was he responding to the train 'feeling' heavier than it was or is there some kind of alarm or video system to keep an eye on the train?

At that specific location they would be looking out for it - it's no accident there was a video camera set up there, to watch the trains but also in part wishing for this to happen. The clue is in the name of the site: "Horseshoe Curve". The track turns more than 180 degrees on an up grade so it is all too easy to pull the train off the track inwards. The radius is well under 200 m, and while the gradient is "only" 1.45% on the curve itself, the drawbar force is determined by the overall rise on the track before it where the train's weight is: that's steeper.

Derailment is such an obvious hazard of this stretch of track that the way it's operated must have been worked out to prevent this happening. But evidently it went wrong this time.

 

[najaB]

On the plus side they're so experienced at clearing derailments the train is already gone and the line appears to be re-opened.

Indeed. One of the first comments on the original video above was something to the effect of "Norfolk Southern must have RJ Corman on speed dial".

They're one of the specialist companies who do rail recoveries in the states.

 

[AndrewE]

At that specific location they would be looking out for it - it's no accident there was a video camera set up there, to watch the trains but also in part wishing for this to happen. The clue is in the name of the site: "Horseshoe Curve". The track turns more than 180 degrees on an up grade so it is all too easy to pull the train off the track inwards. The radius is well under 200 m, and while the gradient is "only" 1.45% on the curve itself, the drawbar force is determined by the overall rise on the track before it where the train's weight is: that's steeper.

Derailment is such an obvious hazard of this stretch of track that the way it's operated must have been worked out to prevent this happening. But evidently it went wrong this time.

Wouldn't helper locos mid train and bankers at the rear have helped here? It seems to me that they just ignore the risk of a train being pulled over on a sharp curve - and we don't know how loaded any of those trucks were. It seems odd to have a huge empty skeletal freight car in the middle of a train (5 min 42 sec) going round such a bend if there was a lot of heavy weight behind it - and it does seem to be the first to go over.

 

[RT4038]

Little of column A, little of column B.

US railroads are not as safe as the UK nor many other networks and have plenty of accidents and derailments (though most are low speed in yards, so not as dramatic as this). Obviously they have the inherent safety of railways which means they are still a safe form of transport in the US but not as safe as they should be. You can play with this site for all sorts of data (make sure to scroll all the way down to check some of the footnotes for context of some of the stats). But for instance in 2025 they had 983 derailments of which 296 were on the mainline so several derailments per week on the mainline.

I think there were a lot more derailments in the UK when we had a far greater percentage of 'loose freight car railroading'. Comparing passenger operations and block load freight/containers with the type of freight railroading the US has is not comparing apples with apples.

 

[ainsworth74]

I think there were a lot more derailments in the UK when we had a far greater percentage of 'loose freight car railroading'. Comparing passenger operations and block load freight/containers with the type of freight railroading the US has is not comparing apples with apples.

I'm not quite sure I get what you mean by "loose freight car railroading"? The US makes more use of auto-couplers than we do for their freight services, they look at our screwlink coupling aghast. If you mean they do more shunting that's true as they still have significant hump yards and wagon load freight alongside train load freight. But I'm not convinced the large number of derailments on the mainline has anything to do with "loose freight cars"? Could you explain?

 

[dgl]

I'm not quite sure I get what you mean by "loose freight car railroading"? The US makes more use of auto-couplers than we do for their freight services, they look at our screwlink coupling aghast. If you mean they do more shunting that's true as they still have significant hump yards and wagon load freight alongside train load freight. But I'm not convinced the large number of derailments on the mainline has anything to do with "loose freight cars"? Could you explain?

I think what they are getting at is you can't compare historical derailment figures as ours would be skewed by unfitted trains having a higher likelihood of derailing and not talking about more recent times with exclusively fitted freight.

 

[ainsworth74]

I think what they are getting at is you can't compare historical derailment figures as ours would be skewed by unfitted trains having a higher likelihood of derailing and not talking about more recent times with exclusively fitted freight.

Oh yes I see. Well I was very much considering now with now rather than historic figures. And I'm still quite sure that the US railroads have far more derailments, even allowing for differences in network size, than we do.

 

[bluegoblin7]

On the afternoon of August 11, 2026, Norfolk Southern manifest freight train 33A, heading westbound, derailed on Horseshoe Curve near Altoona, Pennsylvania. This was caught live on the Virtual Railfan YouTube stream:

Sssh it’s sleeping

 

[Harpo]

How would the driver be aware that there had been a derailment dozens of wagons back on the train?

Three most likely scenarios are, driver notices sudden loss of speed or, extra drag overloads the locos or, brake pipes separate or split, losing brake pressure and applying brakes.

None is guaranteed though as I’ve known miles of track damage from a derailed wheelset (dragging brakes/false flange) that had no impact on progress. But thats a minor ‘off’!

 

[Acathater]

The second box car, the yellow one, looks like the prototype for the Tri-Ang giraffe van!

 

[Rhinojerry]

Wouldn't helper locos mid train and bankers at the rear have helped here? It seems to me that they just ignore the risk of a train being pulled over on a sharp curve - and we don't know how loaded any of those trucks were. It seems odd to have a huge empty skeletal freight car in the middle of a train (5 min 42 sec) going round such a bend if there was a lot of heavy weight behind it - and it does seem to be the first to go over.

That was the cause,empty Centrebeam wagons should be at the rear.Not the first time they have come off and caused a derail.

 

[AdamWW]

Little of column A, little of column B.

US railroads are not as safe as the UK nor many other networks and have plenty of accidents and derailments (though most are low speed in yards, so not as dramatic as this). Obviously they have the inherent safety of railways which means they are still a safe form of transport in the US but not as safe as they should be. You can play with this site for all sorts of data (make sure to scroll all the way down to check some of the footnotes for context of some of the stats). But for instance in 2025 they had 983 derailments of which 296 were on the mainline so several derailments per week on the mainline.

This video:

is worth a watch for anyone who wants to learn a bit more.
It seems from the comments that it's a reasonably fair description - as well as being highly entertaining.

 

[ac6000cw]

Wouldn't helper locos mid train and bankers at the rear have helped here?

Yes. I've visited that area twice, and not seen any mid-train helpers, but rear (and sometimes front as well) manned helpers were commonplace on the heavier trains.

Anyone know if the train in question had helpers on the rear or not?

It seems to me that they just ignore the risk of a train being pulled over on a sharp curve - and we don't know how loaded any of those trucks were.

They don't ignore it - there's rules about placement of empty cars in train consists, particularly for high centre-of-gravity cars like those empty centre-beam lumber (cut timber) cars. It's certainly not a new problem e.g. decades ago Santa Fe (ATSF) spent a lot of money re-engineering the line over the summit of Cajon Pass in California, mainly because they were having too many 'stringlining' incidents around the orginal sharp curve at the summit, as trains had got heavier.

For anyone not familiar with the Horseshoe Curve area, this some of my video from nineteen years ago, taken at the same place (the viewing area) as the Virtual Railfan video - the helper locos back then were pairs of EMD SD40-2/SD40E, nowadays AFAIK pairs of big GEs.

 

[RT4038]

I'm not quite sure I get what you mean by "loose freight car railroading"? The US makes more use of auto-couplers than we do for their freight services, they look at our screwlink coupling aghast. If you mean they do more shunting that's true as they still have significant hump yards and wagon load freight alongside train load freight. But I'm not convinced the large number of derailments on the mainline has anything to do with "loose freight cars"? Could you explain?

What I am meaning is that you need to compare apples with apples. The UK has abandoned freight service with individual rail cars being loaded and despatched, often over great distances, shunted numerous times (often 'on the fly') with the frequent possibility of stress damage, made up into long mixed freight trains and having no 'mother and father' as far as maintenance is concerned. What comparatively little freight trains we have in the UK is block loads, with rakes of cars captive to particular traffic and closely maintained by the same set of people, on lines usually maintained to passenger train speeds/standards. Much of our network carries no (or very little) revenue freight at all, so comparing network size is not reasonable. When the UK had a similar type of individual wagonload freight service, we also had lots of derailments.
What needs to be compared is freight car mileage or even better freight car mileage x tons conveyed, between the two countries, to get a more balanced view.
Safety comes at a cost, which I suspect the US is well aware of and accepts a certain amount of accidents in exchange for a certain level of freight charges - a bit like we do for road transport.

 

[najaB]

When the UK had a similar type of individual wagonload freight service, we also had lots of derailments.

But the counter to that is that back then we didn't have things like wheel defect sensors, or dragging object and hot axle detectors. We also used screw couplings rather than knuckle couplers, and very little of the network was continuous welded rail all of which should significantly reduce the frequency of derailments.

 

[RT4038]

But the counter to that is that back then we didn't have things like wheel defect sensors, or dragging object and hot axle detectors. We also used screw couplings rather than knuckle couplers, and very little of the network was continuous welded rail all of which should significantly reduce the frequency of derailments.

And the trains were slower and there were many, many more people involved in inspecting, checking and operating the trains, but still lots of derailments. Many more wagon movements then.

It wouldn't surprise me if freight car mileage in one day in the US is about equivalent to that in this country for a whole year. The UK has not been immune to some pretty spectacular freight train derailments (Summit Tunnel, Rickerscote and Llangennach, for instance), and those with relatively short block load trains. We have been indeed fortunate that none of these happened in the middle of a town.

Could the US railroads be safer - yes, but as I have said, safety comes at a cost. Do they have more spectacular accidents per ton mile of freight moved? - not so sure about that! It is very difficult to fairly compare, due to vast difference in operating conditions.

 

[GB]

But the counter to that is that back then we didn't have things like wheel defect sensors, or dragging object and hot axle detectors. We also used screw couplings rather than knuckle couplers, and very little of the network was continuous welded rail all of which should significantly reduce the frequency of derailments.

Screw couplings are not an issue. We still use them on the majority of freight today. The benefit (if you can call it that) with screw couplings is they can absorb more twisting motions than knuckle couplers. That's why we don't see many derailments where wagons flip each other over in sequence.

 

[RT4038]

Screw couplings are not an issue. We still use them on the majority of freight today. The benefit (if you can call it that) with screw couplings is they can absorb more twisting motions than knuckle couplers. That's why we don't see many derailments where wagons flip each other over in sequence.

However, British Railways considered the fitting of buck eye couplers to passenger stock as an advancement to help keep passenger stock upright in the event of a derailment.

 

[Taunton]

This type of derailment is sufficiently common in the USA to have its own expression - "String-lining", or pulling the train out straight on curves. Obviously the operators know the criteria to avoid, but can still happen. BNSF (maybe even back to Santa Fe days) spent a considerable amount on realigning and rebuilding The Cajon twisting ascent out of Los Angeles where it was becoming a repeat issue as train lengths grew ever longer. Combinations of superelevation variances, worn flanges, and other aspects contribute. Part-filled tank cars with the contents sloshing on the curve and upsetting the weight balance between left and right can be part of it.

it was something that also caused a lot of development in remote control systems for assisting locomotives like Locotrol, just where to place these in the train to avoid the same issue. "Mid-train Pushers" had long been used to avoid issues with pushing from the rear, but troubles could arise where the neutral point between lead locos pulling and the assisting locos pushing actually passed back behind these rear locos, or alternatively the rear locos were still pushing and compressed the couplers ahead right up when the front locos had throttled back. Alleviations included strain gauges in the assisting locos front coupler to measure what proportion of the load they were taking, it took a good while to understand and get right, and some railroads actually gave up Locotrol and other systems for a while because of this.

 

[ac6000cw]

BNSF (maybe even back to Santa Fe days) spent a considerable amount on realigning and rebuilding The Cajon twisting ascent out of Los Angeles where it was becoming a repeat issue as train lengths grew ever longer.

It was in Santa Fe days (as I mentioned in post #20 above) - the summit was the main problem.

Interestingly, despite having almost continuous back and forth curvature (including the spiral loop and an almost horseshoe curve around Caliente), the Tehachapi Pass line has (AFAIK) never had any significant straightening out - it's basically still on its original (1870s) alignment.

Alleviations included strain gauges in the assisting locos front coupler to measure what proportion of the load they were taking, it took a good while to understand and get right, and some railroads actually gave up Locotrol and other systems for a while because of this.

Also, modern Locotrol allows the lead Engineer (driver) to operate the remote slaves either synchronously (slaves follow the power and brake settings of the lead loco) or async (each slave set is controlled independently via a touch screen interface, allowing better control of the forces/slack along the train in difficult terrain). Some modern locos when used as slaves can also be setup so that when operating syncronously they generate lower tractive efforts than the lead loco, keeping the train stretched out - GE called it 'Controlled Tractive Effort' (CTE) when it was introduced.

 

[43096]

Also, modern Locotrol allows the lead Engineer (driver) to operate the remote slaves either synchronously (slaves follow the power and brake settings of the lead loco) or async (each slave set is controlled independently via a touch screen interface, allowing better control of the forces/slack along the train in difficult terrain). Some modern locos when used as slaves can also be setup so that when operating syncronously they generate lower tractive efforts than the lead loco, keeping the train stretched out - GE called it 'Controlled Tractive Effort' (CTE) when it was introduced.

My understanding is that the CTE equipped locos (e.g. Union Pacifics C44ACCTEs) can reduce their tractive effort by 10% when below 15mph and in DPU (Distributed Power Unit) mode. The decision to use CTE or FTE (Full Tractive Effort) is made during the process to link the lead locos and the DPUs. It is done particularly where the lead locos are DC-motored and the DPU is AC motored to try to avoid the sort of incident this thread is about.

 

[Rhinojerry]

It was in Santa Fe days (as I mentioned in post #20 above) - the summit was the main problem.

Interestingly, despite having almost continuous back and forth curvature (including the spiral loop and an almost horseshoe curve around Caliente), the Tehachapi Pass line has (AFAIK) never had any significant straightening out - it's basically still on its original (1870s) alignment.


Also, modern Locotrol allows the lead Engineer (driver) to operate the remote slaves either synchronously (slaves follow the power and brake settings of the lead loco) or async (each slave set is controlled independently via a touch screen interface, allowing better control of the forces/slack along the train in difficult terrain). Some modern locos when used as slaves can also be setup so that when operating syncronously they generate lower tractive efforts than the lead loco, keeping the train stretched out - GE called it 'Controlled Tractive Effort' (CTE) when it was introduced.

Thats very interesting information,thank you.

 

[Taunton]

A further benefit with modern versions of Distributed Power and Locotrol is the remote units can use their rheostatic braking in combination with those at the front, giving additional mid-train control on descents.