A small, honest experiment with one variable, one notebook, and a year.
In February of 2025, partly out of curiosity and partly out of stubbornness, I started keeping a small notebook on the table beside my bed. Each morning, when I made coffee, I wrote down four things in it: what time I had stopped reading the night before; what book I had been reading; what time I had eventually fallen asleep, to my best estimate; and how I felt at the moment of writing. The whole entry took less than a minute. I kept the practice up for fourteen months. I am, as of this writing, just now stopping.
I want to write about what I found, because the findings are different from what I expected, and because the popular discourse on reading and sleep — which is enormous, and most of which I have read with steadily diminishing patience — does not, in my honest opinion, have very much idea what it is talking about. I have a small data set of one person, fourteen months long, kept with reasonable care. I am not going to claim it generalizes. I am, however, going to claim it taught me something.
What I expected to find
I expected the practice to confirm a version of the standard advice. The standard advice, as it appears in roughly nine of every ten articles on the subject, is that reading before bed is good for sleep, that paper is better than screens, that calmer books work better than tense ones, and that you should stop reading when you feel sleepy and turn out the light promptly. None of this advice is wrong. Most of it, in my data, turned out to be considerably more complicated than the articles suggest.
I also expected — and this was the more personal expectation — that the practice would make me feel guilty about dark romance, which I had been told for years was the wrong kind of bedtime reading. The standard advice strongly implies that you should be reading something gentle and well-paced before sleep. Dark romance is, almost by definition, not gentle and not well-paced. I expected the data to scold me. The data, as it turned out, did not.
What I actually found
Three findings, in roughly the order I noticed them.
The first finding is that the strongest predictor of how well I slept was not what I had been reading. It was what time I had stopped reading. The genre of the book was, in my data, a much weaker signal than the clock. A dark romance finished at 11:30 produced better sleep than a cozy mystery finished at 1:15. The simple variable — when did I close the book — accounted for more of the variance in my sleep quality than every other variable I had been tracking, combined.
This is not a surprising result. It is, however, a result the popular discourse mostly does not emphasize. The discussion of reading and sleep tends to focus on the contents of the reading, because the contents are interesting to argue about. The clock is boring. But the clock turns out to be most of the story. If you want to sleep well, the most consequential thing you can do with your reading is choose, in advance, a stopping time you will respect — and respect it. The book is, in this sense, almost incidental.
The second finding is that the relationship between book intensity and sleep quality, while real, is not the one the standard advice describes. The standard advice says that intense books — thrillers, dark romance, anything with strong tension — keep you awake. My data says that intense books only keep you awake if you read them at the wrong time of day. Read in the early evening, an intense book actually seemed to sleep me better than a calm one. The intensity, processed at a reasonable hour, seemed to help me wear out the kind of restless attention that otherwise turns into late-night thinking. Read in the small hours, the same book did the opposite. The variable, again, was not the book. It was the timing.
The third finding was the one I found most interesting, and the most personal. Over the fourteen months, I started to notice that the books I felt most drawn to read late at night were the books that were, in some way, working on a question I had not yet articulated to myself. The reading was, in part, a way of having that conversation. When the book was helping with the question, I slept well, regardless of how dark or fast-paced the book was. When the book was unrelated to whatever was actually on my mind, I slept badly, regardless of how gentle the book was. The reading was diagnostic, not therapeutic. When I picked the right book for what was actually going on with me, sleep followed. When I picked the wrong book, no amount of cozy prose would help.
What this changed about my reading
The findings, taken together, changed my reading practice in three small ways. I want to describe them, in case any of them are useful.
I now choose a stopping time for the night before I open the book. Not strictly — life is real, and some nights I extend the deadline, and some books are honestly worth the lost sleep. But the default is set in advance. I write it on the bookmark, with a pencil. When I reach the stopping time, I put the book down. I have found, somewhat to my own surprise, that this discipline has not made the reading worse. It has made it, if anything, more deliberate. Knowing the reading is finite makes the time inside the reading feel more present.
I have also gotten better at noticing what I am bringing to the book — what is actually on my mind when I sit down to read. The reading has become a small daily practice in self-honesty, of the kind that the better relationship books recommend without specifying how to actually do it. The "how to actually do it" is, in my experience, partly this: pay attention to which books you reach for, and ask why those, and not the others.
And I have stopped feeling guilty about the dark-romance reading. The data is unambiguous. Dark romance read in the early evening sleeps me well. Dark romance read in the small hours sleeps me poorly. The same is true of literary fiction, of cozy mysteries, of the occasional mediocre thriller I get sucked into despite myself. The genre is not the variable. The genre is, on its own, almost neutral. What matters is when the reading happens, in what state I bring to it, and whether I respect the stopping time. If those three things are right, almost anything sleeps me. If those three things are wrong, almost nothing does.
What the popular discourse gets wrong
I want to spend a paragraph being a little impatient with the popular discourse, which is, after fourteen months of careful attention to my own reading and sleep, the part of my findings I feel most strongly about.
Most articles on reading and sleep treat reading as a pre-sleep activity to be optimized for the singular purpose of producing sleep. They are right that reading helps with sleep, and they are right that some kinds of reading help more than others. But they are wrong about what reading is for. Reading is not, primarily, a sleep aid. It is a practice in its own right, with its own purposes, that happens to also be compatible with sleep when arranged thoughtfully. To treat reading as a sleep optimization tool is to misunderstand the relationship. The reading is the point. The sleep is the consequence of doing the reading well. If you treat the reading as a means to the sleep, you will find yourself reading worse books, less attentively, and sleeping no better.
The right framing, I think, is the one I arrived at slowly through the year. Reading is something I want to be doing. I want to do it well. Doing it well includes — as a kind of ethical obligation to my future self — knowing when to stop. The sleep that follows is not the prize. It is the natural consequence of having read well, and stopped at the right time, and treated the practice as something that deserves my care.
I am stopping the notebook this month, partly because I have learned what it had to teach me and partly because the daily ritual has begun to feel, after fourteen months, like a substitute for the looser noticing it was supposed to support. I will keep the principles. I will lose the form. The notebook itself I am putting on the shelf with the other reading journals, none of which I have ever revisited and all of which I keep anyway. The shelf has an aesthetic at this point. I am pretending that is the reason.