Timing belt Reinstallation direction important?

I’m doing a headgasket job on my 2004 subaru outback 2.5. Imagine that. When I removed the timing belt I didn’t mark a direction on it. Can I reuse it? the belt was changed about 15-20K miles ago. I hate to put on a new belt if I can reuse the old one. If I had marked it to make sure it went on the same way I wouldn’t even question reusing it.

If I already had the old belt off, I would put on a new one. Who knows, it might save you a belt job later, when you sell it.

The hard part is done…It’s minimal cost to replace it compared to the labor that was already done.

If it’s a patch up reinstall the belt. If you want the job done right replace it.

I have a new belt coming that I will probably install, but I’m really wondering what would happen if I used the old belt? I never worry about serpentine belts when i remove them. What makes the timing belt different? I’m on a knowledge quest here more than anything. I agree if I was paying a shop to do this no doubt put on a new belt, but I’m doing it myself so the labors pretty cheap.

The problem is, by the time most anyone has an old belt off, they’ll pretty much always put on a new one, so it’s not like someone’s done it both ways and evaluated the consequences.

The labor involved with replacing a timing belt is considerable and the consequences of a failed timing belt are catastrophic. Would you drop a $4,000 bet on the roll of the dice if the winnings were $5o.?

Some vehicle manufactures do require that the direction of travel is maintained when the timing belt is removed for some reason and then reinstalled.I am not the “Subie guy”, perhaps he will chime in.

$27 for a new one at RockAuto (Dayco or Gates) http://www.rockauto.com/catalog/x,carcode,1440648,parttype,5716

Ok, so the OP ordered the new belt.

I’m still wondering about the actual question. I’m familiar with the recommendation of keeping the direction on reinstall. But I am wondering about the reason for the directionality. So if anyone has a smart answer for that I’d love to hear.


Most timing belts are toothed, unlike a serpentine belt, which has ridges.

A toothed belt has a wear pattern on the leading edge of the tooth over time.
If you were to install the belt backwards, you increase the fatigue on the tooth of the belt, as all the forces that the rubber had become accustomed to over time would then be altered greatly.

I’ve seen many a Ducati timing belt shred itself in short time that had been in use for years, but then reinstalled backwards by accident, during maintenance. Standard practive is that if the belt covers are being removed, that the belts are immediately marked for location (horizontal or vertical cylinder), and direction, to prevent a very expensive Italian engine from being destroyed by accident.


I find that counter-intuitive (rather than simple) - though I’m not disagreeing or arguing the point.

The reason I find it counter-intuitive is that the teeth will have a wear pattern - one side of the tooth will take more wear & stress. Which to me means that you should reverse it once in while to spread the wear & stress. You obviously can’t compare it to tire rotation, but, well…kind of like rotating a tire.

Presumably intuition is not a good guide in this case.

I am just spit-balling here, but my recumbent bicycle came with directional tires. Don’t directional tires have less rolling friction than ordinary tires? Perhaps it is the same with the belt. If the belt is toothed, and I am pretty sure it is, losing traction isn’t likely to be an issue, so perhaps they make timing belts to they provide a little less resistance going in one direction than the other.

I looked carefully with a magnifying glass at the tooth edges of two old VW timing belts with 60,000 miles on each of them. At first I thought that one tooth edge appeared to be a little smoother than the opposite edge but after a few minutes of looking, I can not see enough of a significant difference in an effort to identify the tooth edge most highly loaded. That tooth edge would be the side pushed by the crankshaft pulley. The crank pulley is has the smallest diameter and has the least number of teeth engaged with the belt, at least with my VW making the tooth loading highest at the crank pulley.

A timing belt on a Ducati may not be a parallel situation to use for direct comparison. The Ducati that I am familiar with is air cooled and the belts are in very close proximity to the cylinder head and cylinder, making the Ducati belts more highly thermally stressed than the belt in a water cooled engine.

You might look at the tooth edges on your Subaru belt to see if one edge is more highly polished than the other. The belt would then travel in a direction so that the less polished edge leads the more polished tooth edge. That is my guess with a little reasoning behind it.

I can’t find anything on the back side of my used belts to identify travel direction but possibly I don’t know what to look for or else need more magnification.

I’m in agreement with Bladecutter about belt orientation and considering it’s an interfernece engine a new belt would be a given and 25-30 bucks is pretty cheap peace of mind.

There’s a few questions and points that could be made.
What about the belt tensioners and water pump?

You’re doing a head gasket job. Checked the heads for flatness? Subaru heads are very short in length and logic would dictate that warpage would be near impossible. However, a non-scientific guesstimate of the Subaru heads I’ve checked showed about 80% of them exceeded the maximum .002 of an inch allowed. Two or less will pull down with the head bolts.

You should use spray Copper Coat on the head gaskets, torque the head bolts, run the engine for a few hundred miles, and then recheck the torque on those head bolts.

Also use care if checking and adjusting the valve lash. Hopefully none of these tips are after the fact… :slight_smile:

Compare a new belt to a used belt:

A new belt fits into the lugs on the pulleys nice and snug.
A used one not nearly as snug.

If you reverse the belt, wear will be quicker on the opposite side, and will make the belt even looser even quicker. This will lead to valves that are no longer exactly where they should be in comparison to crankshaft position.

A loose belt is going to be generating more heat, which will wear out the rubber edge even quicker. Eventually, that leading edge will get worn to the point that is might be able to jump timing easily.

The purpose of using the square blocks is to help prevent the timing from jumping teeth for the longest period of time possible.

Again, using the Ducati timing belt analogy:

Back when the cam belt adjustment and replacement specs were initial adjustment at 500 miles, then adjust again every 3k miles, and replace at the 2 year or 12k mile mark, the belts had round teeth.

When they changed to adjustment specs of every 6k miles, but still kept the 12k mile/2 year replacement point, they switched to square teeth on the belts.

They have since revised the adjustment point now to 7500, with replacement at 15k miles, but still every 2 years for a time constraint.

These are for both air cooled engines, and liquid cooled engines, so the thermal envelope doesn’t change the application any. Another thing is that the tensioners are manually set, and are not automatically set. Imagine having to adjust the tension on your car’s belt every 7500 miles. Uggh.


I think the problem would be that when you run the belt, you’re stretching the material that makes up the belt in one direction, then when you reverse it, you’re stretching it the other way, which tends to weaken it more.