Sterols of 31 samples of green coffee beans of the arabica and robusta varieties have been analysed by gas chromatography flame ionization detector. The lipids were Soxhlet extracted from ground coffee beans into hexane. The extract was evaporated and the residue was treated with 0.2% of 5α-cholestane-3β-ol (internal standard). The lipids were saponified and the sterolic fraction of the unsaponifiable part was separated by thin layer chromatography. Sterols were treated with a silylating reagent to convert them into trimethyl silyl derivatives. This analysis was performed on a column (30 m×0.32 mm i.d.) coated with a bonded stationary phase HP-5 (5% diphenyl–95% methylpolysiloxane; 0.2 μm) with hydrogen (0.7 ml min⁻¹) as a carrier gas, isothermal temperature at 265°C and flame ionization detector. By using the sterols as chemical descriptors pattern recognition techniques were applied to differentiate between arabica and robusta green coffee varieties. Δ⁵avenasterol and sitostanol were found to be the most differentiating variables.