When the microsphere method of blood flow measurement is used in small organs, the number of spheres actually recovered may be small. This introduces errors which, however, are not necessarily important as compared with the biological variation. A mathematical model was constructed which allowed estimation of the extra uncertainty caused by the low number of spheres. It was applied to an experimental material of over 100 measurements of cochlear blood flow in the rat, where only about 60 spheres were recovered per cochlea. It was concluded that this paucity of spheres introduced only a small to moderate extra error. Simple approximate formulae were derived allowing easy estimation of the uncertainty caused by low number of spheres under different conditions. In general, there appear to be many situations where fewer microspheres than are commonly used would suffice.