How many papers can a PhD student realistically read in a day?

The reason this question causes so much anxiety is that the visible evidence is misleading. You see colleagues citing enormous literatures and assume they read all of it at depth. They did not. They triaged most of it, read a fraction closely, and built the citation list from a mixture of the two, which is standard practice rather than a shortcut.

Deep reading is genuinely slow. A dense empirical paper in an unfamiliar area, read to the point where you could explain the design and criticize the analysis, takes two to four hours. Two of those in a day is a good day, and a third is usually worse than the first two because attention has gone.

If you want a defensible daily target rather than a boast, take one paper. Méndez recommends reserving a daily reading slot for at least one paper, and explicitly allows you to finish it across several short sessions instead of one block, while rejecting cursory reading driven by a deadline (Méndez, 2018, p. 1). One consequential paper understood per day, plus lighter scans when the day allows, is the workload rule the evidence actually supports.

The survey numbers do not give you a quota either. Hubbard and Dunbar report an estimate that scientists spend 23% of their working time reading, and that the number of papers read per year rose from 188 in 1993 to 280 in 2005 while total reading time barely moved (Hubbard and Dunbar, 2017, pp. 1-2). That is a story about selectivity, not speed. Their career-stage comparison is more useful to you: 41% of PhD students reported reading papers they had found themselves more than once a week, against 63% of academics (Hubbard and Dunbar, 2017, pp. 7-8). Academics who report several papers a day are not the doctoral benchmark.

Skimming is fast and it is not lesser. A five-minute structured pass reliably tells you what a paper claims, what it measured, and whether it matters to you. Fifteen of those in a day is achievable and is how most people cover a field.

The number that actually matters is neither of these. It is how many usable claims you extracted, which is what your chapter is built from. Some days one paper yields six. Some days twelve papers yield nothing, which is also information about where you are looking.

How thoroughly am I supposed to be reading?

It depends on the paper's role in your argument. Works your contribution rests on, or that you intend to argue against, need close reading. Works that supply one fact need the passage containing that fact. Reading everything closely is not thoroughness, it is a failure to prioritize.

A useful sorting question: if this paper turned out to be wrong, what would happen to my thesis? If the answer is that a chapter collapses, read it properly, check its methods, and look for replications. If the answer is that one sentence needs a different citation, read the relevant section and move on.

Working researchers already sort this way. In Hubbard and Dunbar's survey, 68% recommended reading selectively within a paper rather than front to back, and 53% named practical tactics such as using selected sections to decide whether a paper deserves fuller attention, rereading, taking time, and not getting stuck in technical detail (Hubbard and Dunbar, 2017, p. 11). Carey, Steiner, and Petri make the same point as a rule about purpose: a newcomer to a technique may go through every figure panel, someone entering a new field may read only the introduction and conclusion, and someone following a particular author may skim the whole article (Carey et al., 2020, p. 2). Two or three papers in a day is realistic when the purpose is triage or orientation. It is not the same act as understanding two or three unfamiliar studies.

The papers people under-read are the ones they agree with. Confirming sources get a glance while contradicting ones get scrutiny, and that asymmetry is worth correcting deliberately, because a supportive study with a fatal flaw is the one an examiner will find.

Is it normal to not finish most papers I open?

Yes, and experienced researchers do it more, not less. The skill you are building is deciding quickly whether a paper answers your question, and the correct response to most papers is to stop. Abandoning a paper after four minutes with a reason recorded is a completed action.

The guilt attached to this is a leftover from coursework, where finishing the reading was the task. In research the task is answering a question, and papers are instruments. Nobody feels guilty about not reading a whole reference manual.

Nor does difficulty go away with seniority. 42% of the academics Hubbard and Dunbar surveyed described reading papers as frustrating even while calling it valuable, and the authors argue that PhD students need structured support with research reading because an undergraduate degree does not reliably produce it (Hubbard and Dunbar, 2017, p. 13). If a paper takes you two hours and leaves you irritated, that is the reported experience of people two decades ahead of you, not a sign that you are slow.

What does deserve attention is the pattern of abandonment. If you consistently stop at the methods section because the statistics are unfamiliar, that is a skill gap worth closing rather than a triage decision, and it is better to notice it now than in a viva.