The performances of two laser Doppler flow meters (Periflux model PF3 and Moor Blood Flow Monitor model MBF3D) were investigated. Recordings were made while diluted blood was pumped at different rates through the pulp cavities of extracted human or pig teeth. The probe of each instrument was fixed to the enamel surface 2 mm from the original position of the gingival margin. Both instruments performed similarly, although the Moor gave the better signal/noise ratios. Both were capable of detecting flow in the core of the pulp as well as the superficial part closest to the probe tip. With either instrument, and with any one dilution of blood (range 0.5-45% v/v red cells), there was a near-linear relationship between the blood-flow signal and the flow rate of blood through the tooth. However, when data obtained with different red-cell concentrations were compared, a good correlation between the blood-flow signal and red-cell flux (product of concentration and mean velocity) was obtained only with red-cell concentrations up to 1% v/v. Therefore these instruments would give an accurate indication of blood flow changes only under conditions in which either this value was not exceeded or if the red-cell volume fraction remained constant; neither of which can be assumed to apply when recording from teeth in situ. The signals representing the concentration of moving blood cells were unreliable in both instruments. It is concluded that the information provided by these laser Doppler flow meters can be ambiguous and must be interpreted with care.