The theory of the electromagnetic blood flow measuring technique is extended from the well known case (conductive liquid flowing through an insulating tube) to more realistic situations. First the conductivity of the vessel is taken into account, and the electric potentials in both liquid and vessel wall are calculated. The potential difference V between two points on the outside of the vessel and on an axis at right angles to both magnetic field B and the flow v is computed. The comparison is made with the classical flowmeter result V = 2Ba[UNK] (a = inner radius of vessel, [UNK] = mean flow velocity). For an average artery, with a ratio of inside diameter to outside diameter of 0.85, the error is found to be in the order of -7 per cent. The blood is assumed to be four times as conductive as the wall tissue. The induced potentials are then calculated in the liquid, in the vessel wall, and in a thin liquid conductive layer surrounding the artery. A film of serous fluid which is likely to exist between a blood vessel and the applied flowmeter sleeve creates an additional shunt. The voltage between the flowmeter electrodes deviates from the expected result by -10 to -15 per cent if the film thickness is 3 per cent of the outside radius of the tube. The evidence is therefore established that flowmeter cuffs should fit the blood vessels accurately to minimize errors.