How long should an annotation be?

Whatever your instructor says, and if they say nothing, one paragraph. The widely repeated 100 to 200 word rule does not come from any style manual. APA's Publication Manual covers annotated bibliographies in Section 9.51 and gives formatting only, with no length. The MLA Style Center gives no word count either, saying annotations usually do not exceed one paragraph. Purdue OWL explicitly declines to set one and tells you to consider the purpose and your instructor's directions. So the only number that binds is the one in your assignment brief, and in its absence a paragraph that summarises, evaluates and connects the source is the right length by construction, which in practice lands somewhere between 100 and 200 words.

What is the difference between a summary and an evaluation in an annotation?

A summary says what the source contains. An evaluation says how much weight it can bear. This is the single most common way an annotated bibliography loses marks: the student writes a competent paragraph describing the study, and the assignment asked what it was worth. Evaluation means judging the evidence, not praising the paper: what kind of study is it, how big and how selected was the sample, what did the authors themselves say it cannot show, and what would have to be true for the finding to transfer to your case. If every sentence in your annotation could have been written by someone reading only the abstract, you have written a summary.

How do you format an annotated bibliography in APA, MLA and Chicago?

The citation is formatted exactly as it would be in a normal reference list in each style, and only the annotation's placement differs. APA puts the annotation in a new paragraph below the reference, with the whole annotation indented half an inch like a block quotation and the first line not indented further. The MLA Style Center's guidance is to add a line space after the entry and begin the annotation with a paragraph indent, and to title the list Annotated Bibliography or Annotated List of Works Cited. Chicago is the most permissive: its guidance allows an annotation to follow the publication details directly or to start a new line, usually indented from the left margin.

Is an annotated bibliography the same as a literature review?

No, and the difference is synthesis. An annotated bibliography keeps each source in its own box: one entry, one self-contained annotation, and no source is compared with another. A literature review is continuous prose organised by theme, where the point of a paragraph is what several sources together establish or dispute. The bibliography is usually the groundwork you do before writing the review, which is why it is set as an assignment: it forces you to read and judge each source separately before you have to make them talk to each other.

How to write an annotated bibliography (three examples)

The CiteOwl Team· Updated 11 August 2026· 12 min read

Every guide to annotated bibliographies hands you a template with square brackets in it, and a template is exactly what you do not need. The hard part is not the shape of the paragraph. It is knowing what an evaluation sounds like when it is not a summary wearing a hat. So this page gives you three finished annotations on three real papers, one each in APA, MLA and Chicago, the same source annotated badly and then well so the difference is visible, and what the style manuals actually mandate about length, which is far less than the internet claims.

Contents

What the assignment is testing
The length rule no style manual contains
What evaluation actually means
The same source, summarised and evaluated
A finished entry in APA
A finished entry in MLA
A finished entry in Chicago
The formatting differences that are real
Writing one without faking the reading

An annotated bibliography is a list of sources where every entry carries the full citation and then a short paragraph, the annotation, about the source. A plain bibliography proves you found something. An annotated one proves you read it and formed a view.

What the assignment is testing

It is set as a milestone toward a longer paper or a literature review, and it exists to make you do the judging early, while there is still time to drop a source that does not hold up. That is also what it is marked on. The citations are worth a few marks and take ten minutes to get right. The annotations are worth the rest, and the gap between a good one and a mediocre one is entirely whether you took a position on the source.

It tests one other thing, in a way most assignments cannot. You can fake a summary from an abstract in four minutes. You cannot fake an evaluation, because evaluating a study means knowing what its sample was and what its authors admitted it could not show, and neither is in the abstract.

The length rule no style manual contains

Search for annotation length and you will find "100 to 200 words", or "about 150 words", repeated across hundreds of pages. Here is where that comes from: nowhere. We checked the manuals.

APA. The Publication Manual covers annotated bibliographies in Section 9.51, and the guidance is formatting only: order the references alphabetically as in a reference list, put each annotation in a new paragraph below its reference, and indent the whole annotation half an inch, as you would a block quotation. No word count anywhere (APA Style, reference lists versus bibliographies).
MLA. The MLA Style Center says annotations usually do not exceed one paragraph, that they are generally written as succinct phrases though complete sentences are fine, and that they should "not rehash minor details, cite evidence, quote the author, or recount steps in an argument" (MLA Style Center on annotated bibliographies). No word count.
Chicago. Its guidance on annotated bibliographies allows the annotation to follow the publication details directly, sometimes in brackets where only a few entries are annotated, or to start a new line, usually indented from the left margin. No word count.
Purdue OWL, where most students actually land, sets no length either, and tells you to consider the purpose of your bibliography and your instructor's directions when deciding how much to include (Purdue OWL, annotated bibliography samples).

So the only binding number is the one in your assignment brief. Absent that, write the paragraph that does the three jobs below and let it come out the length it comes out. The three finished examples on this page were written that way, without counting, and came out at 152, 155 and 152 words, which is presumably how the convention got invented in the first place.

Our opinion, since this page is otherwise full of other people's rules: the 150-word convention has done more harm than good, because a word count is the easiest requirement in an assignment to satisfy and students satisfy it with summary, which is the cheapest prose available. If your instructor gives a number, treat it as a ceiling rather than a target, and spend the last forty words on the judgement rather than on three more sentences about what the study did.

What evaluation actually means

A complete annotation does three things: it summarises the source, evaluates it, and says how it fits your project. Purdue OWL describes the same three moves and notes that an annotation may do one, two or all three depending on what the assignment wants. Nearly everyone can do the first. The second is where the marks are, and it is not a paragraph of praise.

Evaluation answers four questions, one sentence each.

What kind of evidence is this? An experiment, a survey, a model, a meta-analysis, a case study, an argument. This decides what the source can establish at all.
How much of it, and how was it selected? Sample size and how it was chosen, or, for a model, what it was calibrated against.
What do the authors say it cannot show? Take it from their own limitations paragraph. This is the sentence that proves you read past the abstract.
What would have to be true for it to apply to your case? Different population, different decade, different setting. Naming the gap is not a weakness in your annotation, it is the analysis.

Words that signal real evaluation: because, although, only, which means, does not establish, applies where. Words that signal summary in disguise: important, valuable, insightful, comprehensive. If a sentence in your annotation calls a study "a valuable contribution to the field", delete it. It carries no information.

The same source, summarised and evaluated

One real paper, annotated twice. The source is a randomised experiment on working from home, published in The Quarterly Journal of Economics in 2015.

What most students hand in. This study examines whether working from home affects employee performance. The researchers ran an experiment at a large Chinese travel company, Ctrip, in which call centre employees worked from home for nine months. They found that home working led to a 13% increase in performance, and that employees reported higher work satisfaction while attrition fell by half. The authors conclude that working from home can benefit both employees and firms. This source provides useful information about remote work and will be helpful for my paper on the topic.

Every sentence there is true, and the paragraph is worth very little. It could have been written from the abstract, it takes no position, and the last sentence is filler. Now the same source, evaluated.

What the assignment wanted. Bloom and colleagues randomly assigned call centre staff at Ctrip, a 16,000-employee Chinese travel agency, to work either from home or in the office for nine months, and measured a 13% performance gain, of which 9 percentage points came from working more minutes per shift and 4 from more calls per minute. Attrition halved and satisfaction rose, but promotion rates conditional on performance fell. The randomisation is what makes this worth citing: nearly every other study in this literature compares people who chose to work from home with people who did not, and this one does not have to. Its limits are just as sharp. One firm, one highly measurable job, volunteers only, and a period before 2020, when home working was rare rather than default. It supports a claim about individual output in routine measurable work, and cannot support one about collaborative or creative work, which is exactly where my paper has to be careful.

The second version takes a position, defends it in the same breath, names what the paper cannot support, and closes by telling the reader what the writer will do about that. It costs 67 words more than the first and is worth several times as much, and every one of those words came from having opened the paper.

A finished entry in APA

APA reference formatting, sentence case in the article title, annotation in a new paragraph below and indented half an inch as a block.

Freeman, S., Eddy, S. L., McDonough, M., Smith, M. K., Okoroafor, N., Jordt, H., & Wenderoth, M. P. (2014). Active learning increases student performance in science, engineering, and mathematics. Proceedings of the National Academy of Sciences, 111(23), 8410–8415. https://doi.org/10.1073/pnas.1319030111

This meta-analysis pooled 225 studies comparing student performance in undergraduate science, technology, engineering and mathematics courses under traditional lecturing against active learning. Across 158 studies, examination and concept-inventory scores rose by 0.47 standard deviations under active learning, about a 6% improvement, and across 67 studies the odds of failing were 1.95 times higher under lecturing. Scale is the strength here: the effect held across every STEM discipline examined, was largest in classes of 50 or fewer, and trim-and-fill analysis suggested it is not an artefact of publication bias. The weakness is what the label covers, since "active learning" pools a wide range of classroom practices into one category, so the finding supports abandoning pure lecturing rather than adopting any particular technique. For my paper on lecture format and attainment this is the anchor citation for the overall effect, and I will pair it with a single-course study before saying anything about mechanism.

A finished entry in MLA

MLA formatting, title case, "et al." for three or more authors, annotation after a line space with a paragraph indent.

Bloom, Nicholas, et al. "Does Working from Home Work? Evidence from a Chinese Experiment." The Quarterly Journal of Economics, vol. 130, no. 1, 2015, pp. 165–218. https://doi.org/10.1093/qje/qju032.

Reports a nine-month randomised experiment in which call centre employees at Ctrip, a Chinese travel agency of 16,000 staff, were assigned to work either from home or in the office. Performance rose 13%, split between more minutes worked per shift and more calls handled per minute; attrition halved; promotion rates conditional on performance fell. When the firm later let the experimental group choose for themselves, more than half switched and the measured gains nearly doubled, to 22%. The random assignment is unusual and valuable in this literature, where most evidence compares self-selected home workers against office workers. Generalisation is the limit: one firm, one routine and highly measurable role, volunteers only, and a pre-2020 setting in which home working was the exception rather than the norm. I use it as the strongest causal evidence available for routine work, and as the clearest illustration of why the post-2020 survey literature cannot be read the same way.

A finished entry in Chicago

Chicago bibliography style, all authors listed, annotation on a new line and indented.

Lebreton, Laurent C. M., Joost van der Zwet, Jan-Willem Damsteeg, Boyan Slat, Anthony Andrady, and Julia Reisser. "River Plastic Emissions to the World's Oceans." Nature Communications 8 (2017): 15611. https://doi.org/10.1038/ncomms15611.

Builds a global model of plastic entering the ocean from rivers, driven by waste management data, population density and hydrology, and calibrated against the measurements available in the published literature. It estimates that between 1.15 and 2.41 million tonnes of plastic waste enter the ocean from rivers each year, that over 74% of it moves between May and October, and that the twenty most polluting rivers, mostly in Asia, account for 67% of the global total. This is a modelling study rather than a measurement campaign and the calibration data are thin, which matters, because a later analysis built on different assumptions attributed 80% of riverine emissions to more than a thousand rivers rather than to twenty. I cite it as the origin of the concentrated-sources claim and put the competing estimate directly beside it, since the disagreement between the two is the part of my paper that has something to argue.

The formatting differences that are real

The annotation itself reads the same in every style. Only the citation and the indentation change.

Element	APA 7	MLA 9	Chicago 17
Article title	Sentence case	Title case, in quotation marks	Title case, in quotation marks
Authors	All up to 20, ampersand before the last	First author, then "et al." for three or more	All listed, "and" before the last
Where the year sits	After the authors, in brackets	After the volume and issue	After the volume, in brackets
Annotation placement	New paragraph below, whole block indented half an inch	Line space, then a paragraph indent	New line, usually indented, or straight after the details
List title	Annotated Bibliography	Annotated Bibliography or Annotated List of Works Cited	Bibliography, or a descriptive title

Two notes worth having. Library guides commonly tell you to indent an MLA annotation a full inch, following the sample in the Handbook, while the MLA Style Center's own guidance describes a line space and a paragraph indent. Either is defensible, so pick one, stay consistent, and follow your department where it has a preference. And in APA you do not need to cite the work inside its own annotation, since the reference above it makes the source obvious, though you should cite any other work you bring in.

Writing one without faking the reading

Work source by source, in this order: find it, read it, cite it, annotate it. The reading is the assignment. What you need from each paper is exactly the four evaluation questions above, and our guide to reading a research paper has a fifteen-minute routine that gets all four without reading the whole thing.

The shortcut, asking a chatbot for a list of sources with ready-made summaries, fails here more visibly than anywhere else, because an annotation has to contain specifics that exist only in the real paper. A general model will produce a fluent 150-word annotation of a study that was never written, and a marker who opens one reference has your whole list. CiteOwl works the other way round: it searches real databases, reads the paper, and shows the passage behind anything it tells you about it, so the source you are annotating is one you can open yourself. The judgement and the words stay yours, which is the part being marked. To check a list you already have, our free citation checker resolves each entry against CrossRef, and the citation generator turns a DOI into a formatted reference in APA, MLA or Chicago with no account.