Caffeine is the most consumed active stimulant. About 80% of pregnant women consume caffeine orally on daily basis. Many reports indicated consumption of >200 mg caffeine during pregnancy could increase the likelihood of miscarriage. In this paper, we developed a pregnancy physiological based pharmacokinetic/pharmacodynamic (PBPK/PD) model for caffeine to examine association between maternal caffeine consumption during pregnancy and caffeine plasma levels at doses lower and higher than 200 mg to predict changes in caffeine concentrations across the 3 trimesters, and to predict associated changes in caffeine PD parameters. Two models were successfully developed using GastroPlus™ software, a non-pregnant model for validation purposes and a pregnant model for validation and prediction of maternal caffeine plasma concentrations following single and multiple dosing. Using observed and predicted data, we were able to validate and simulate PK changes of caffeine in non-pregnant women and the PD effect of caffeine on certain enzymes and catecholamines associated with caffeine intake. Furthermore, the pregnancy PBPK model successfully predicted changes in caffeine PK across the three trimesters. Caffeine increased exposure during pregnancy was related to reduced activity of caffeine metabolizing enzyme CYP1A2. The model also predicted increased levels of caffeine in the fetoplacental compartment (FPC) due to increased maternal caffeine plasma concentrations. Increased caffeine levels in maternal blood was accompanied with greater inhibition of phosphodiesterase enzyme, higher cyclic adenosine monophosphate and greater increase of epinephrine levels, which could increase risk of pregnancy loss. The application of the developed PBPK model to predict PD effect could provide a useful tool to help define potential cut-offs for caffeine intake in various stages of pregnancy.