Roast level moves the caffeine in your cup by only a few percent either way, and whether light or dark comes out ahead depends on how you are measuring it: by weight of beans, by scoop, or by what actually ends up dissolved in the brew. A 2024 study in Scientific Reports, run out of Berry College and Drexel University, found something sharper than either half of the usual argument: how you brew the coffee moves the caffeine more than how dark you roasted it.
Does a Lighter Roast Really Hold More Caffeine?
Sometimes, and only under one specific condition. Royal Coffee, the green importer that supplied the beans for the Berry/Drexel study, lays out the honest version: "with identical brewing conditions, lighter roasts have more caffeine in the resulting brew," but flip the comparison to identical extraction percentages instead of identical brew method, and "a coffee brewed with darker roasts has more caffeine." Those are not two sources disagreeing. They are the same data answering two different questions, and almost everyone who repeats "light roast has more caffeine" as a flat rule is quoting only the first half.
Royal calls the real relationship "complicated," and that is the correct word for it, not a hedge. Two mechanisms pull in opposite directions at once: roasting drives some caffeine out of the bean, and roasting also opens the bean's structure so the brewer can pull more of whatever caffeine is left into the water. Which effect wins depends on exactly how far you roasted, exactly how you brewed, and exactly what you are counting.
What Actually Happens to Caffeine Inside the Roaster?
Caffeine is a stubborn molecule. It does not break down easily at roasting temperatures, which is the part most casual explanations skip past on their way to a verdict. Royal Coffee's reporting on the Berry/Drexel data is specific about when it finally starts to matter: "sublimation effects did not become significant until the bean probe temperature exceeded values of approximately 400 to 420°F." That is well past where most light and medium roasts are pulled. Below that threshold, caffeine mostly just sits there while almost everything else in the bean is changing around it: sugars caramelizing, chlorogenic acids breaking down, water and mass leaving as steam and gas.
That mass loss is the real lever, and it is bigger than people expect. The study, covered by Daily Coffee News, roasted two Ethiopian green coffees, one washed and one natural, to five separate roast degrees on an Aillio Bullet, then rested each batch ten days before grinding, sieving for a consistent particle size, and brewing with an inverted AeroPress at 212°F and a 15:1 water-to-coffee ratio, with the coffee mass held constant across every sample. Caffeine and chlorogenic acid were measured directly by liquid chromatography, not estimated. A darker roast loses more mass to that process, mostly moisture and CO2 rather than caffeine, which means a given weight of dark roast beans actually started out as more green coffee, and therefore more caffeine-bearing material, than the same weight of a lighter roast. Weigh your dose on a scale, as most specialty brewing already recommends, and some of the old "light roast has more caffeine" folk wisdom quietly reverses itself before you have brewed a drop.
Which Roast Actually Produced the Most Caffeine in the Cup?
Neither end. The Berry/Drexel study's cleanest finding is the one almost nobody repeats: "maximum caffeine levels were observed in cups of medium roasts that experienced about 14 to 16 percent loss of mass during roasting." Not light, not dark: medium, sitting at the point where the bean has opened up enough to extract well but has not yet lost so much soluble material that there is less caffeine left to pull. Chlorogenic acids, used in the study as a chemical marker of roast degree, dropped steadily and predictably as the roasts got darker, which the researchers note matches earlier work. Caffeine did not follow nearly as tidy a curve, because it was competing against extraction yield the whole time rather than simply degrading on a schedule.
Daily Coffee News's framing of the study's own conclusion is the one worth remembering over any rule of thumb: "brewing matters more than roasting." Extraction yield, how much soluble material actually left the grounds and ended up in your cup, moved the measured caffeine more than roast degree did on its own. An under-extracted light roast and an over-extracted dark roast can land caffeine levels closer together than the roast colors would suggest, and a careless brew of either one can out- or under-perform a careful brew of the other on caffeine alone.
A separate, smaller test backs up the same conclusion from the opposite direction. Swiss Water Process's own Director of Coffee, Mike Strumpf, roasted a single coffee to three roast levels at five different roasting speeds and measured the caffeine left in the beans afterward. He found no pattern at all: "no consistent trend emerged to show the level correlated with how much caffeine ended up in the coffee bean itself," and his own summary calls the results "consistently inconsistent." Between a controlled university study finding a real but modest medium-roast peak and a working decaffeination company's in-house test finding no pattern whatsoever, the one thing both results agree on is that roast color is a poor predictor of caffeine on its own.
So Why Does the "Light Roast Has More Caffeine" Rule Keep Getting Repeated?
Because it is a half-truth stated as a whole one, and half-truths are stickier than qualified ones. "Lighter roasts have more caffeine in the resulting brew" is a real finding under real conditions, and Royal Coffee's own data backs it, under identical brewing. The sentence just quietly drops its own condition on the way into a blog post, a cafe menu, or a roaster's bag copy, and a real, bounded result turns into a flat claim it was never meant to carry. We have published that same flattened version ourselves, on an older page comparing light and medium roasts, and it is wrong for the same reason every other version of it is wrong: it states the light-roast half of Royal's finding as a universal rule and never mentions extraction yield, brew method, or the medium-roast peak the actual study measured. That correction is overdue on our end, and this article is where it belongs.
Does Inheritance's Own Shelf Settle Any of This?
Not as far as dark roast goes, and I want to say that plainly rather than let an empty shelf answer the question on its own. We roast light to medium across the board. Checked against live inventory the day this was written: one light roast, four medium, and zero dark among the coffees we currently have on the shelf. If you wanted to put the "dark roast has less caffeine, pound for pound" half of this debate to your own test at home with one of our bags, you cannot — we are simply the wrong roaster to settle that particular comparison for you, because we have never stocked a dark roast to compare against.
What our shelf can speak to is the medium-roast half of the finding, because most of what we roast sits close to where the Berry/Drexel study measured its caffeine peak. ASCENT by Inheritance, our own medium-roast espresso blend, is built for exactly the kind of careful, consistent extraction the study's own conclusion points to as the bigger lever. We roast it to order every Wednesday in Pensacola and grind it to your brewer at no charge, and the caffeine in your cup from it will owe far more to your brew ratio and your extraction than to the roast level on the bag.
Chase caffeine through your brewing, not your roast level. ASCENT by Inheritance is our medium-roast espresso blend, roasted to order every Wednesday and ground to your brewer at no charge.
