It depends to the journal, there are some journals accept similarity index less than 25% and some other journals don't accept similarity index more than15%...so you have to check the journal requirements...generally, 15% is acceptable in a wide range of journals
And plagiarism can bite back, not only in offending copyright laws, but also in copying wrong information -- which becomes yours. Right information, on the other hand, exists, and is coherently attributable to the author, sooner or later, so plagiarism fails there too.
Always write what you can prove, above doubt, and never copy without attribution. That way, you can control errors to a small value, as small as you please, with drafts and time, to be useful to you and others.
Note, though, that duplicate thought exists. Maxwell's equations in electromagnetism were already covariant with special relativity before Einstein was even born!
Accepting plagiarism will bite them back too. See http://www.collective-evolution.com/2015/05/16/editor-in-chief-of-worlds-best-known-medical-journal-half-of-all-the-literature-is-false/
It depends on one's nature of research and it is very subjective as the existing measurement tool that available is only based on the similarity check (Turnitin). For my humble opinion , only the experts on their particular field of interests will verify either the paper is leading to plagarism.
And the result of any poor professionalism accepted here, shows.
Marcia Angell, a physician and longtime Editor in Chief of the New England Medical Journal (NEMJ), which is considered to be one of the most prestigious peer-reviewed medical journals in the world, makes her view of the subject quite plain:
“It is simply no longer possible to believe much of the clinical research that is published, or to rely on the judgment of trusted physicians or authoritative medical guidelines. I take no pleasure in this conclusion, which I reached slowly and reluctantly over my two decades as an editor of the New England Journal of Medicine”