We recently introduced an imaging application combining dynamic positron emission tomography (PET), a modification of the two-tissue compartment model, and constrained parameter estimation. The objective of this method, which we have called neurotransmitter PET (ntPET), is to estimate the timing of neurotransmitter (NT) kinetics. The time course of NT release in response to drugs or other stimuli may provide information about the function of the brain. In this paper, we evaluate two alternate formulations of ntPET, one which uses arterial blood samples as the plasma input function (ART) and another which uses a reference region-derived approximation of the plasma input (REF). Simulated data with strong (moderate noise, prominent NT response) and weak (high noise, subtle NT response) signals were analyzed with ntPET using ART and REF. Both methods were able to recover NT profiles resembling the true response, with temporal resolution better than 1 min for strong signals and 3 min for weak signals. Despite potential disadvantages, REF yielded results rivaling those of the ART method. When a sufficiently robust response is anticipated and knowledge of absolute timing is not necessary, the REF method is an appropriate alternative to ART, which is more demanding experimentally.