Coffee roast levels without false precision
Light, medium, and dark sound like fixed points. They are not. Roast level describes how far a coffee was roasted, but the boundaries and traditional names change between sellers, machines, and colour systems.
The useful goal is not deciding whether a batch is City+ or Full City. It is describing the result consistently enough to repeat it, then deciding whether you like it in the cup.
Reviewed 9 September 2026
Use several clues together
Colour is obvious, but lighting, processing, and the difference between surface and centre colour affect what you see. First and second crack add context, although some coffees crack softly and neither event defines a universal boundary.
Displayed temperature is most useful within one machine. Probe placement, thickness, airflow, and batch size can shift the reading. Weight loss adds another comparison, but initial moisture and roast path affect it. Use colour, cracks, time, machine readings, weight loss, and tasting together.
- Light
- Usually brown and dry, often with more evident acidity and origin character. Light does not automatically mean underdeveloped.
- Medium
- A wide middle region with many possible balances of acidity, sweetness, body, bitterness, and roast character.
- Dark
- Commonly associated with progression into second crack, more smoke and bitterness, and a greater chance of surface oil. Exact boundaries vary.
Development ratio is not a score
Development time is the interval from your chosen start of first crack to the end. Dividing it by total roast time gives a development-time ratio. It can help repeat similar batches, but it cannot diagnose underdevelopment, baking, or balance by itself.
The often-repeated 15 to 25 percent range is a roasting convention, largely associated with drum profiles. It is not a scientifically validated quality window. Two roasts with the same ratio can have different heat application, total time, end point, internal colour, and flavour. Fast air roasts make universal percentages especially misleading.
Roast level and caffeine
Caffeine is relatively stable in ordinary roasting, although measured content varies with species, coffee, roast, dose, and brewing. Roasting also reduces bean mass and increases volume, so comparisons by scoop, bean count, and roasted weight answer different questions.
Studies do not all produce the same light-versus-dark result, but roast colour is a poor caffeine-control tool. Weigh your dose. To reduce caffeine substantially, use less coffee or choose decaffeinated coffee.
Choose by tasting
Seller notes can suggest a direction. Coffee bought for floral or fruit character is often explored lighter; coffee chosen for body, chocolate, or milk drinks may invite a darker approach. These are possibilities, not rules attached to origins.
Roast the same lot twice with a clear difference. Brew on comparable days with the same recipe. Decide which is sweeter, clearer, more balanced, or better suited to your drink. That result matters more than the label.
Common questions
What are the main roast levels?
Light, medium, medium-dark, and dark are widely understood groups. Cinnamon, City, Full City, Vienna, and French are traditional names, but their boundaries are not standardised.
What temperature defines a medium roast?
No universal probe temperature does. Establish a useful range on your machine by pairing its readings with colour, crack observations, weight loss, and tasting.
Does 20 percent development guarantee a good roast?
No. The ratio omits heat application, total time, end point, bean properties, and sensory quality. Use it to compare similar batches, then let tasting decide.
Sources and further reading
These references support the technical claims above. Practical recommendations are identified as starting points because coffees and equipment differ.
Put this batch to work.
Percolate connects green stock, roast observations, blends, and tastings. Use it when you want the last cup beside you while you plan the next roast.
Start roasting