MRG Effitas Consumer Assessment & Certification Programme 03/2026

[Parkinsond]

https://www.mrg-effitas.com/wp-content/uploads/2026/04/MRG_Effitas_Consumer_Assessmnet_03_2026.pdf

 

[Khushal]

View attachment 297092

https://www.mrg-effitas.com/wp-content/uploads/2026/04/MRG_Effitas_Consumer_Assessmnet_03_2026.pdf

Outstanding Results by Malwarebytes.
As expected sub par results by Gdata, Mcafee and Sophos. Gdata's dual engine got defeated by BD's single engine.

 

[Miraculix]

McAfee's results were surprisingly poor.

 

[SeriousHoax]

Outstanding Results by Malwarebytes.
As expected sub par results by Gdata, Mcafee and Sophos. Gdata's dual engine got defeated by BD's single engine.

Yeah, BD's behavior blocker shone more than G-Data's in this case.

Malwarebytes tends to do well in mrg-effitas's test compared to others, which is a bit surprising to me.

 

[Parkinsond]

Malwarebytes tends to do well in mrg-effitas's test compared to others, which is a bit surprising to me.

The as Avast/AVG/Norton in AVC tests.

Gdata's dual engine got defeated by BD's single engine.

Dual engine without B powerful behavioral analysis.

McAfee's results were surprisingly poor.

McAfee, MD, Panda; all are cloud-based, all take time to react.

 

[7Oz-64]

Alls security products are on default settings, these results should be slightly different with non-default settings and i think most users tweaks security solutions.

 

[Khushal]

View attachment 297092

https://www.mrg-effitas.com/wp-content/uploads/2026/04/MRG_Effitas_Consumer_Assessmnet_03_2026.pdf

https://forums.malwarebytes.com/topic/334863-zkm-stealer-2600-java-ratinfostealer/ Meanwhile malwarebytes unable to detect Java malware which has been reported to FBI

 

[Sorrento]

Oh NO! All the Av's I like are useless, formatting now, starting again, cannot get my head around to using Mbam though so I disagree with the findings

 

[Parkinsond]

Oh NO! All the Av's I like are useless, formatting now, starting again, cannot get my head around to using Mbam though so I disagree with the findings

I do not believe MBAM is good; I use either MD or AVG after removing all component including web protection, just leaving file shield, behavioral shield, and ransomware shield.

 

[Khushal]

Oh NO! All the Av's I like are useless, formatting now, starting again, cannot get my head around to using Mbam though so I disagree with the findings

That is why consistency is key. Bitdefender and Kaspersky have made such a solid reputation with their strong showing in thousands of independent tests that one anomaly can be discounted. Meanwhile Malwarebytes 100 percent detection here seems like an anomaly itself.

 

[Sorrento]

'One swallow does not a summer make' - As with all things in life, AV's & tests need to be taken in aggregate, & decisions made for our own needs, one size don't ft all I reckon, interesting though - If someone asks I will give my opinion as to how to set a system up including router internet type OS/ etc but I never feel that I can say this is what you should use because it works for me, ask 20 users on here the ideal way you might get 20 different replies & all are correct at least for them.

 

[Andy Ful]

Interesting Real-World test, especially when we look at missed samples.
MCAfee .................... 50 / 300
G Data ...................... 22 / 300
Sophos ..................... 19 / 300
Norton ..................... 4 / 300
Avast ........................ 4 / 300

In the standard Real-World tests, most AVs miss only a few samples per 300 total samples, and by using statistical methods, those that miss 0-3 samples must usually be awarded (Level 1 certificate).

The criteria of the Level 1 certificate seem to be artificial (statistically unsupported).
The Level 2 certificate looks reasonable (can be supported by statistics).

 

[Parkinsond]

Interesting Real-World test, especially when we look at missed samples.
MCAfee .................... 50 / 300
G Data ...................... 22 / 300
Sophos ..................... 19 / 300
Norton ..................... 4 / 300
Avast ........................ 4 / 300

In the standard Real-World tests, most AVs miss only a few samples per 300 total samples, and by using statistical methods, those that miss 0-3 samples must usually be awarded (Level 1 certificate).

The criteria of the Level 1 certificate seem to be artificial (statistically unsupported).
The Level 2 certificate looks reasonable (can be supported by statistics).

The only thing I find interesting regarding MRG Effitas tests is displaying behavioral-related protection, similar to AVLab tests which display the post-execution detection rate.

 

[Khushal]

The only thing I find interesting regarding MRG Effitas tests is displaying behavioral-related protection, similar to AVLab tests which display the post-execution detection rate.

No i find something even more pleasing about them.

If an ITW ransomware or wiper sample executes successfully and results in permanent loss of user files or irreversible system damage, the product cannot be
certified, regardless of its overall block rate.

 

[Parkinsond]

No i find something even more pleasing about them.

AVLab has one more advantage; it reports time to remediation which can make a difference in mitigating the file encryption and/or data exfiltration, regardless of post-execution detection by behavioral analysis.