How to Make Light Roast Espresso, According to Chemists (2025)

Time was, in those forgotten years called the "late ’90s,” espresso only meant dark roast. And “espresso roast” meant oily-dark beans. At least that’s what they still write on Starbucks bags.

But then came the third-wave coffee revolution of the early noughties, dedicated to the proposition that all beans aren’t created equal. A new generation of coffee roasters and baristas began to question the notion that coffee beans should be roasted into submission, preferring to highlight agricultural origins and fruity aromas by roasting lighter. Often much, much lighter.

For espresso, the results weren’t always good.

The first time I was offered a truly light roast espresso was, of course, in Portland, Oregon. It was an airy, postindustrial café of the sort that was popular 15 years ago, decorated with rough-hewn wood and unrelated machine parts—decor that’s meant to showcase not just blue-collar grit but the very idea of process. This fruit-forward, Ethiopian espresso didn't just spring into being: It was made, labored upon, with real human hands against the drum.

Anyway, I hated it. Sure, my little double shot was languidly syrupy and drenched in swirling burnt-caramel crema, like a lot of fancy espresso. It was also intense and astringent—bracingly aromatic but also bitter and pucker-sour and seemingly determined to suck all spittle from my tongue. If this was the big new thing, I was pretty sure I didn't want it.

I mention this because if you drink a lot of espresso, I'm pretty sure you've had this same battery-acid cup. And if you have, you quite possibly came away with the notion that espresso was never meant to be made with lighter-roast coffee. Who would trade all that roundness and chocolatey depth for … whatever that is?

The answer, 15 years later, is me.

I'm a little late to the party, perhaps. But very light roast espresso is my biggest rediscovery of the past couple years—a taste I cultivated especially while testing two or more espresso machines each month over the past year to assess the best. I’ll often pull 50 or 100 shots on each machine, which leaves plenty of time for experimentation.