Italian GWOT Steel: the Beretta AR-70/90

While the Italian military did adopt the AR-70, it did not actually issue them to all troops. Most continued to use the 7.62mm BM-59 until 1990 when the Beretta AR-70/90 was adopted. This rifle was a substantial rework and improvement of the AR-70, using AR-pattern magazines and a 1:7″ twist barrel to run the SS109 62gr ammunition chosen as the new NATO standard in 1981. The 70/90 also added a 2-position gas regulator to the design, while carrying over the folding bipod and grenade launching capabilities of its predecessor.

The AR-70/90 was made in semiautomatic from by Beretta for the civilian market, but only after US regulation prohibited its importation. As a result, these rifles were essentially nonexistent in the US until a few years ago, when a number of cut-up parts kits were imported. Several small shops set up to make them into semiautomatic rifles with new US-made receivers, including Brimstone Arms, who made this one for me.


  1. I completely missed where Ian removed the flash hider assembly, I think he left that step on the cutting room floor.

  2. It’s sorta amazing how Beretta and SIG were collaborating on a roller-delay locked system together, couldn’t make that work, and then moved to a pair of very similar AK-esque rotary bolt rifles.

    It’s almost as if 5.56mm in roller-delay is hard to do, or something. HK is the only company to actually pull that off, and everyone that’s quit paying for their TDP and manufacturing assistance has moved off 5.56mm in that system…

    What doesn’t work, doesn’t work. Unless, apparently, you’re the gnomes of the Black Forest working for HK. Too bad they all retired, and that nobody apparently captured their “secret sauce” for making those systems reliably and affordably.

      • The AMELI is one of the examples of failure I was thinking of… That cute little thing looks like a 5.56mm MG42, but at it’s heart, it’s not roller-locked, it’s roller-delayed. And, they never quite got it working well enough to be affordable to produce.

    • It seems that it works, but with constraints, even for H&K.
      In general, with lever/roller delay, the greater the rate of reduction of bolt weight you want, the pickier the system becomes. With .223 Rem / 5.56 NATO the system seems to become barely usable exactly at the weight needed for a workable AR. See the FAMAS F1, that needed the specific French ammos, or the HK33, that’s also ammo-specific. If you want to reliably use a different ammo, or even a different barrel lenght, you need a different bolt.
      In the Beretta/SIG collaboration, SIG guys were the experts in roller delaying, due to the PE-57. I can picture the Beretta guys, once heard of all those limitations with the .223 Rem, stating “ok, but the AK action works ALWAYS! AS IT IS!”

      • FAMAS was lever-delay, not roller.

        The root issue is, I believe, the high pressures and the curves at which they develop for 5.56mm. Roller-delay works really well for something like 7.92X33; less well for 7.62X51, and apparently not at all well for 5.56mm cartridges. There are reasons the Swiss and the Italians abandoned their development programs for that cartridge in that mechanism, and moved to the rotary bolt for it.

        I’d speculate that there are equally poor cartridges for the AK system, but they’re such edge cases that nobody has really run into one. Remember, the AK is basically a rationalized Garand mechanism turned upside down… Which itself is an apparent adaptation of a French semi-auto design. Nothing new under the sun, I fear.

        • Infact I wrote “lever/roller delay”.
          Because the principle is the same. You exchange mass of the bolt (required for a pure blowback action) for speed of a part of it.
          And the problems are the same. The greater the weight reduction you want of the bolt, the pickier the system becomes.
          And yes, some charateristic of the .223 / 5.56 seems to be particularly unfavourable to roller delay. As some charateristic of the 5.7 FN seems to be particularly unfavourable to browning short recoil actions (while it works for the equally bottleneck and even higher pressure 7.5 FK).
          In the aforementioned CETME Ameli, the roller delay worked. And it worked because it has an heavy bolt with little delay. What didn’t work that well was the weapon around the rollers. That seemed to have some of the problems you experienced with another MG I don’t want to nominate (IE the MG tending to self-disassemble in heavy use).

        • The only major problem with the AK system is feed geometry. It’s happiest with cartridge cases with a pronounced taper. So it works very well with 7.62 x 39 or 5.45 x 39.5, adequately with 7.62 x 54R, reasonably with 7.9 x 57, and tends to have problems with 7.62 x 51 or 5.56 x 45.

          As for the roller-lock, it works best with pressure levels and curves that don’t stray too far from those of the original WW2 7.9 x 57 Wehrmacht spec. Hence it works OK with the original 7.62 x 51 NATO, but not the reduced-charge versions everybody came up with in a vain attempt to make full-auto fire with 7.62 NATO rifles practical.

          This probably explains why everybody, even the French, have settled on some version or other of MAG 58 in 7.62 x 51. The basic Browning gas-operated system is more tolerant of variances in ammunition quality and powder characteristics.

          It would be kind of interesting to take an MAG of some sort, an MG42/59 or other ’42 variant, and a Browning M1917 or ’19, all in 7.62 x 51 NATO and all on their proper tripods, out to a range with a truckload of ammunition and see which one went the longest on MRBS and MRBF.

          I suspect all of them would outlast most HK G3 based “LMGs”.

          Of course, they’d all outperform any M60 version.



        • @Dogwalker,

          You’re right; I missed the “lever” bit on your post on my first hasty read-through.

          I haven’t got any real “stick time” on an AMELI, just having talked to a bunch of guys who did, and having handled a couple of examples. I don’t know how well the bolt and the action did, compared to the rest of the gun, but everyone who fired one or was carrying one universally described it as a lovely concept with lousy execution.


          I’d love to have the money to do something like that, but barring a win of the Lotto, ain’t happening. I think that the HK MG offerings all suffered from their “Let’s make all the things roller-delay!!!!”

          There’s probably an HK roller-delay facecream out there. Somewhere.

          There’s almost always going to be a problem when you try to apply a specific system absolutely everywhere. Toggle-locked was great in the lever-actions of the time, but when you try to build a friggin’ submachinegun around the concept?

          There is such a thing as taking it all way too far. Another case in point? Extending the Martini action to build an LMG… Madsen, I’m looking at you.

  3. Quick edit here, the small spring clip on the front takedown pin should be under the tab on the other side of the rifle. This allows it to stay in place for standard bolt maintenance when the upper is simply pivoted from the lower, but when the full upper/lower need to be separated, the pin is retained by the spring clip onto the upper assembly. That way both spring clip and pin wouldn’t be so easily lost. Both front and rear takedown pins should be captive.

  4. Since rifle-grenade launching was such an important feature of the AR-70/90, maybe its ‘carry-handle’ was intended for more than a close-range sighting system? Similar to how the carry-handle of the FAMAS was supposed to assist rifle-grenade launching for indirect-fire? Rifle-grenade Indirect-fire might also be why the butt of the AR-70/90 has a thick rubber buttplate?

  5. You can see the FN influence in the FAL pistol grip and the FN-49 gas piston flutes. I bet it pained Beretta deeply to have to make a gun that ugly, though.

Leave a Reply

Your email address will not be published.