Guarana (Paullinia cupana) is an Amazonian native fruit, whose seeds are rich in caffeine and phenolic compounds. Due to the economic importance of guarana, it is necessary to improve the knowledge of its chemical characteristics, which may vary depending on several factors, such as soil characteristics, crop variety, and environmental conditions. In this context, this study describes the application of quantitative nuclear magnetic resonance associated with principal components analysis to evaluate the composition of guarana seeds from the two largest Brazilian producer regions: Amazonas and Bahia states. The principal component analysis discriminated the samples according to the source, revealing that the seeds from Bahia contain higher caffeine and phenolic contents (catechin, epi-catechin), but the seeds from Amazonas present higher content of fatty acids. Additionally, caffeine was quantified using quantitative nuclear magnetic resonance and liquid chromatography coupled to photodiode array detector, which corroborate the chemometrics. The results show a slightly larger content of caffeine for seeds from Bahia. Therefore, qNMR combined with chemometrics was important to detect the metabolic variability of guarana seeds from different regions.