Despite the fact that around 20000 cups of coffee per second are produced worldwide (making coffee the second most-traded commodity in the world), coffee extraction is not well understood yet. This Engineering Doctorate Thesis seeks to advance the fundamental engineering understanding of coffee extraction. This aim is based on the current need of industry to optimise soluble coffee process (as stress on water and energy is increasing), and the growing popularity of On-Demand coffee systems. The macrostructure, microstructure and extraction parameters of roast and ground coffee were investigated. The findings from this study were used in a multi-scale extraction model that portrays the extraction of coffee soluble solids as the combination of phenomena taking place at the particle scale (~μm), and the packed bed scale (~cm). Effective diffusion coefficients in the range of 10\(^-\)\(^1\)\(^1\) m\(^2\) s\(^-\)\(^1\) were shown to offer the better fit to experimental data if a single effective diffusion coefficient is to be used. The model was shown to predict literature extraction data for caffeine, chlorogenic acids and trigonelline in espresso coffees. A new methodology to estimate the permeability of roast and ground coffee in steady state was also developed. Permeability values resulted to lie between 10\(^-\)\(^1\)\(^3\)-10\(^-\)\(^1\)\(^4\) m\(^2\).