The aims of this article are to review the properties of noise in magnitude MR images to clarify the terminology used when referring to the noise and to discourage the use of the terms Rician noise and Rician noise bias. The distribution of measured MR pixel intensities in the presence of noise is known to be Rician, and the width of this distribution is directly related to the Gaussian noise on the measured real and imaginary signals. It is the pixel magnitude values that follow the Rician distribution, not the noise. The term Rician noise should be used cautiously or, better still, avoided completely since inherent to this terminology is behavior that is not normally associated with noise, such as dependence on signal strength. This terminology is misleading and can lead to conceptual and practical misunderstandings. It is better to relate the image noise to the Gaussian noise on the real and imaginary signals. (C) 2008 Wiley Periodicals. Inc. Concepts Magn Reson Part A 32A: 409-416, 2008.A.C-B. and I.C. thank the Ontario Research and Development Challenge Fund for financial support, the Ottawa Health Research Institute. and Dr. Greo O. Cron. C.T. thanks his grant sponsor Fondecyt (#1060008) for their support.8FONDEFicameron@ottawahospital.on.caOntario Research and Development Challenge Fund; Ottawa Health Research Institute; Fondecyt [1060008]6FONDEF32