The reference tissue model allows for quantification of receptor kinetics without measuring the arterial input function, thus avoiding arterial cannulation and time-consuming metabolite measurements. The model contains four parameters, of which the binding potential (BP) is the parameter of interest. Although BP is robust, convergence rates are slow and the other parameters can have large standard errors. To overcome this problem, a simplified reference tissue containing only three parameters was developed. This new three-parameter model was compared with the previous four-parameter model using a variety of PET studies: [11C]SCH 23390 (D1 receptor) and [11C]raclopride (D2 receptor) in humans, and [11C]SCH 23390, [11C]raclopride and [11C]RTI-121 (dopamine transporter) in rats. The BP values obtained from both models were essentially the same for all cases. In addition, the three-parameter model was insensitive to starting values, produced stable results for the other parameters (small standard errors), and converged rapidly. In conclusion, for the ligands tested the three-parameter model is a better choice, combining increased convergence rate with increased stability.