At Wikipedia, "throwing the baby out with the bathwater" refers to edits which remove additional information beyond the scope of a valid deletion, as well as to rationales at discussion pages that extend a core content policies rationale beyond its scope of validity. Reversion of other editors' work should be done only when necessary, and as a matter of policy must be done sparingly.

Types of baby/bathwater actions and positions include:

A general misunderstanding of the Help:Reverting (WP:REVERT) process and Wikipedia:Reverting (WP:RV) essay as meaning "undo anything that is not perfect":

An editor determines a portion of another contributor's edit needs to be deleted, but removes the entire contribution instead of just the portion that needs to be deleted.

• Example: "Edit summary: A sentence had a grammatical error, and one claim lacks a cite, so I'm just reverting the whole lot. I don't have time to clean up those problems in your 30K addition."

Some more specific cases are detailed below, involving misapprehensions of particular policies and guidelines.

Misinterpretation of the Wikipedia:Identifying reliable sources (WP:RS) guideline:

For a source that is unreliable regarding a particular topic, an editor deletes all content that is based on the source—including information for which the source is reliable.

• Example: In a European court case, The Destruction of Dresden author was determined to have misrepresented World War II casualty numbers that were difficult to historically quantify because of the number of refugees in the city. Deleting Wikipedia content based on the book's details about other topics, such as allied bombing operations (numbers of sortie aircraft, etc.) which were substantially researched by the author, is a form of "baby and bathwater" editing.

Failure to understand the Wikipedia:No original research § Primary, secondary and tertiary sources (WP:PSTS) policy:

A revert-minded editor argues that any time a source contains an opinion or new idea that it is a primary source, always, for everything it contains, in every situation.

• Example: "This news article concludes with an anecdote by the journalist, so we can't trust anything it says."
• Example: "This newspaper publishes all kinds of biased op-eds, so all of its articles are suspect."
• Example: "This systematic review is worthless, because it has a section speculating about ongoing research the results of which could affect its analysis."
• Policy: WP:PSTS defines sources as primary, secondary, or tertiary with regard to their content- and context-specific value. It does not in any way definite any particular author, publisher, genre of publication, or medium as categorically only one of these types at all times. A large number of individual sources are primary, secondary, and tertiary all at the same, for different claims in different context and depending on what material in them is being cited.

Often inverted: This kind of argument is often seen in reversed form, wherein an editor who wants to revert a revert, to restore something they did not properly source, argues that all examples of a type of source, or all content from a particular source, "must" be reliable and secondary for one thing because it is for others. This is a different kind of baby–bathwater situation: All objections, no matter how well-reasoned, are tossed out based on the editor's faith in the reputation of the publisher or writer.

• Example: "What do you mean this op-ed in the New York Times by the Religious Foundation for Moral Reform isn't a reliable source? Are you calling the New York Times unreliable?!"
• Example: "Of course this theoretical paper on a possible approach to cold fusion is a reliable secondary source. The author is one of the most respected physicists in the world."
• Example: "It's absurd to say the World Health Organization isn't a reliable source in this article. It doesn't matter that it's an article about particle physics. Notable is notable." (Extra demerits for confusing a publisher's notability with its reputability; even the world's worst tabloids are often notable.)
• Reality: News sources are not uniformly reliable across all topics or with regard to all of their content. Entertainment, religious and political topics, for example, often are accused of having either reportorial biases or editorial biases, depending on the source and writer. Op-eds, advice columns, movie reviews, and other material that expresses an individual's opinion are primary not secondary source material. An article by a sports journalist may be reliable for sport statistics but not for the scientific details of physics or biochemistry.
• Reality: Academic journals are not uniformly secondary sources. Most of them contain a great deal of, or may even specialize in, new and unverified research findings, which are primary sources for their data and conclusions, and cannot be secondary for anything other than literature review material given as the background for the research.
• Reality: Authorities on one thing are not authorities on everything, and they are not authoritative even in their sphere of influence when they are just offering their own personal opinion, organizational political stance, or new and unverified hypothesis.

Overreaction Wikipedia:Biographies of living people (WP:BLP) policy:

An editor concludes that a source does not meet Wikipedia's stringent policy regarding content about living people, but deletes content that is not about the living person but just happens to be in their article, or which is cited to some other, reliable source.

• Policy: WP:BLP requires that any unsourced or poorly sourced material about living persons may be removed immediately and without waiting for a consensus discussion, if it could be contentious in any way (negative, positive, or simply dubious).

Incorrectly applying Wikipedia:Neutral point of view (WP:NPOV) policy:

Content from one editor is determined to have a portion that has a non-neutral point of view ("PoV") or which gives undue weight to a source which does, but the editor's entire contribution(s) to the page, including unbiased information with credible citations, is/are deleted under the auspices of the NPOV policy.

• Example: "That author was in favor of the conservatives' position on deregulating this industry; reverting everything sourced to his books."
• Example: "I'm reverting all these paragraphs by that IP editor; the phrasing, near the end, about the prosecution is clearly advancing a personal bias."

Unawareness of Wikipedia:Verifiability § Accessibility (WP:SOURCEACCESS) policy:

An editor determines content has a citation to a source work that is out of print, not available to the public for free, in a foreign language, only exists offline, is in an obscure medium, requires registration to access, or is otherwise inconvenient to verify, and deletes the citation and/or the material relying on it.

• Policy: These are never valid reasons for such a deletion (though a source that does not exist any longer, and cannot be recovered by any means at any expense, is not a valid source under policy, since it cannot be verified at all).

False belief that the WP:Verifiability (WP:V) policy requires citations, much less particular citation formatting, for everything:

• Example: "Reverting addition of the year in which this riot took place since there's no source cited."
• Example: "Undid addition to discography section; no inline citation."
• Example: "I reverted your messy citations; please see WP:CITE for how to format them properly."
• Policy: WP:V requires inline citations for all quotations and for any material challenged or likely to be challenged, not for all content, and the citations do not otherwise have to be formatted a certain way or be any more detailed than is necessary to identify the source. All content must be verifiable with reliable sources, but an uncontroversial statement is not actually required to be verified with a citation at all unless and until it is controverted. (But see the living persons exception above.)

Any questionable claim should be examined on its own merits, and no source found to meet WP:RS requirements – or any content based on such a source – should be summarily "thrown out" without examining its validity under in the context under applicable Wikipedia policies and guidelines, closely and carefully. If, and only if, the claim is clearly unsupportable by sources that are independent, reliable, and secondary for the matter in question, should it be simply removed on the basis that the source is "not reliable enough" or "not secondary" (though it may need to be removed for some other policy-based reason). And "too much trouble to fish out of other edits" is never a reasonable revert rationale, unless the other edits were also demonstrably problematic. An editor with an eye to reverting has a responsibility to make sure the baby is not in the bathwater.

• WP:Three-revert rule (policy)
• WP:Reverting (essay), including section advising to revert only the parts of an edit you disagree with
• WP:Restoring part of a reverted edit
• WP:Alternatives to reversion (essay)
• WP:Revert only when necessary (essay)
• Help:Reversion (how-to)