The effect of critical surface tension on the initial retention of microorganisms from unstimulated human saliva was tested in a flow cell system. Prior to each experiment the total numbers and the morphotypes of microorganisms present in saliva were recorded. The test surfaces were prepared to display known increasing critical surface tensions, as verified and standardized by contact angle measurements. Surfaces of initially low (20-22 mN/m), medium (35-38 mN/m) and high (greater than 50 mN/m) critical surface tension were exposed to saliva at a flow rate of 1 ml/min. Microbiota and biofilm material associated with the test surface after 15 min of salivary exposure, were then subjected to standard detachment forces, by introducing a cell-free rinsing fluid at two different shear rates. Both the attachment and the detachment phases were executed at room temperature or 37 degrees C. The retained population was counted in three different zones of the test surfaces with a light microscope and statistically tested for correlation to the main variables (critical surface tension, flow rate and temperature) and interactions. Retention success was significantly dependent on the initial critical surface tension and the flow rate. Surfaces of medium critical surface tension, representative of human tooth surfaces and most restorative dental materials, retained the highest numbers of microorganisms in comparison with the other surfaces tested, with no statistically verified selectivity in proportions of retained coccoid and rodshaped microorganisms for any surface. A 30-fold increase of the flow rate resulted in a 70-80% reduction of the retention success, with a higher relative number of cocci present on all the test surfaces. These results demonstrate that initial retention of microorganisms to surfaces is non-specific with regard to morphotypes, but is strongly related both to the mechanical removal forces and the surface energetic state of the solid surface exposed. Retention of microbial populations at interfaces might, therefore, be controllable by advance selection of the critical surface tensions and predicted if shear forces at given sites are known.