Positron emission tomography (PET) is used in drug development to assist dose selection and to establish the relationship between blood and tissue pharmacokinetics (PKs). We present a new biomathematical approach that allows prediction of repeat-dose (RD) brain target occupancy (TO) using occupancy data obtained after administration of a single dose (SD). A PET study incorporating a sequential adaptive design was conducted in 10 healthy male adults who underwent 4 PET scans with [(11)C]DASB ([(11)C]N,N-dimethyl-2-(2-amino-4-cyanophenylthio) benzylamine): 1 at baseline, 2 after 20 mg SD of the 5-hydroxytryptamine transporter (5-HTT) inhibitor duloxetine, and 1 after 4 days daily administration of 20 mg duloxetine. An adaptive design was used to select optimal times after SD for measurement of occupancy. Both direct and indirect PK/TO models were fitted to the SD data to characterise the model parameters and then applied to a predicted RD duloxetine plasma time course to predict the 5-HTT occupancy after RD. Repeat-dose prediction from the indirect model (OC(50)=2.62±0.93 ng/mL) was significantly better (P<0.05) than that from the direct model (OC(50)=2.29±1.11 ng/mL). This approach increases the value of SD occupancy studies that are performed as part of first time in human drug development programmes by providing an estimate of the dose required to achieve the desired TO at RD.