This paper is concerned mainly with artifacts and errors which are encountered in the measurement of blood flow by means of a magnetically induced voltage. The history of the method is reviewed, together with an outline of the theory. The effect of the magnet dimensions on currents induced in the blood stream is discussed. A full account is given of the electrical properties of metallic electrodes immersed in an electrolyte, and their importance in reducing interference in a flowmeter of the alternating current type is explained. The properties of tubular screens in an electrolyte are discussed. A description is given of motor-electrolytic and motor-thermal potentials, and how these artifacts are influenced by the design of the electrodes and input circuit of the amplifier. The ways in which unwanted voltages arise in phase quadrature with the desired flow voltage are described, and the necessity for accurate construction of the flowmeter magnet and electrode assembly is emphasized. A method is suggested for reducing the required accuracy of assembly in the axial direction, and the importance of phase-sensitive detection of the induced alternating flow voltage is emphasized. The electrode system and amplifier are examined in relation to the signal/thermal noise ratio, which (apart from quadrature component and artifacts) determines the limit of sensitivity of a given flowmeter. Hydrodynamic and electromagnetic pressure losses are briefly discussed. The essential requirements are given for distortionless amplification and detection of the flow voltage, together with an outline of the apparatus which has been found useful for the experimental verification of calculated flowmeter characteristics. Finally, some examples are given of potential sources of error in previous work using magnetic flowmeters.