Areca nut (AN) is a carcinogen; its chewing cessation is, therefore, of worldwide interest. However, cessation biomarkers are lacking. We sought to establish arecoline in chewers' buccal cells (BCs) as a biomarker for AN dose. Self-reported AN doses, expressed as the average AN load ("AANL"), the product of AN amount, chewing time, and chewing frequency, were correlated by regression analysis with chewers' BC arecoline, measured by liquid chromatography mass spectrometry. We then determined whether associations differed between Class 1 chewers (who consume AN alone or with slaked lime, leaf, and/or spices) and Class 2 chewers (who consume any combination of the aforementioned ingredients plus tobacco). Among the 103 chewers, 28 Class 1 and 39 Class 2 chewers had detectable arecoline levels, which were used for analyses. A linear regression of cube-root transformed AANL on equally transformed BC arecoline levels provided the best model fit; resulting slopes and corresponding correlation coefficients were 0.86 and 0.40 (p< 0.01) for all; 1.09 and 0.51 (p< 0.01) for Class 1 chewers; 0.35 and 0.17 (p= 0.29) for Class 2 chewers; and 0.94 and 0.45 (p< 0.01), and 0.79 and 0.37 (p= 0.08), respectively, for those who included or excluded lime. Relationships between AANL and BC arecoline levels were similar between chewers who included or excluded lime (p= 0.76), but less between chewing classes (p= 0.14). This provides confidence that BC arecoline can generally act as a reliable biomarker for AN dose, useful for estimating efficacy in AN cessation studies and population-based chewing assessments.