Partial parallel imaging (PPI) techniques using array coils and multichannel receivers have become an effective approach to achieving fast magnetic resonance imaging (MRI). This article presents a Matlab toolbox called PULSAR (Parallel imaging Utilizing Localized Surface-coil Acquisition and Reconstruction) that can simulate the data acquisition and image reconstruction, and analyze performance of five common PPI techniques. PULSAR can simulate sensitivity functions of rectangular loop coils using a quasi-static model based on Biot-Savart's Law and undersampled multichannel data acquisition on a rectilinear k-space grid. In addition, PULSAR provides performance evaluation of the techniques based on artifact power (AP), signal-to-noise ratio (SNR), and computational complexity. In this article, the structure and functionality of the PULSAR toolbox are described. Examples using both the simulated and real four-channel and eight-channel data were used to demonstrate the utilities of the toolbox and to show that PULSAR is a convenient and effective means to study the PPI under different coil geometries, acquisition strategies, and reconstruction methods. © 2007 Wiley Periodicals, Inc. Concepts Magn Reson Part B (Magn Reson Engineering) 31B: 24–36, 2007