This study aimed to investigate the impact of methylxanthine abstinence (MA) periods on CYP1A2 activity in individuals with varying levels of caffeine consumption through development of a population pharmacokinetic model of caffeine and its major metabolite paraxanthine. This study developed and evaluated a mixed-effects pharmacokinetic model for caffeine and paraxanthine concentration-time data derived from a sequential single-dose cross-over study in healthy male volunteers (n = 30) who received oral 100 mg caffeine doses. Participants received caffeine with and without a MA period. Participants were classified as low (0-100 mg/d), medium (100-200 mg/d), or high (>200 mg/d) caffeine consumers (LCCs, MCCs, or HCCs, respectively). All caffeine and paraxanthine concentration-time data were simultaneously modeled. Caffeine pharmacokinetics was described by a two-compartment model with first-order absorption and two first-order elimination pathways. Paraxanthine was described by a one-compartment model with first-order absorption and elimination. Among LCCs (n = 16) and MCCs (n = 9), there was no difference in the mean (95% confidence interval) total apparent caffeine clearance (CL) between the MA period [LCCs: 6.88 (5.61-8.16 l/h); MCCs: 10.09 (7.57-12.60 l/h)] versus the no MA period [LCCs: 6.22 (4.97-7.46 l/h); MCCs: 9.68 (7.12-12.24 l/h)]. The mean CL among HCCs (n = 5) was considerably higher in the MA period [10.48 (5.62-15.33 l/h)] compared with the no MA period [6.30 (3.40-9.20 l/h)] (P < 0.05). The decrease in CL in the no MA period among HCC appears to be due to alternative caffeine elimination pathways, rather than CYP1A2.