Abstract Using the MicroDrop((R)) microdispenser system, a novel approach for high throughput cultivation assays for the determination of numbers of culturable bacteria, and for the isolation of bacteria in liquid media was established. The MicroDrop device works similar to an ink jet printer. Droplets of 150-200 pl are created at the nozzle of a glass micropipette by means of a computer-driven piezo transducer, and are dispensed automatically at predetermined positions with the aid of a XYZ-positioning system. The actual drop volume is highly reproducible and is determined by the pulse duration, the pulse frequency and the micropipette geometry. Culture media in 96-well microtiter plates were inoculated with constant numbers of bacteria from three different natural freshwater lakes. The number of culturable bacteria in the sample can be calculated from the frequency of wells showing bacterial growth, based on a binomial distribution of culturable cells. Our method was compared to the conventional most probable number (MPN) approach, the technique presently most often used for the determination of bacterial culturability and for the isolation of numerically dominant culturable bacteria. As opposed to the MPN technique, our approach yields data with much higher statistical significance (i.e. a 10 times lower standard deviation) due to the higher number of parallels which can be performed in each microtiter plate. The values of culturable bacteria as determined by the MPN and MicroDrop techniques were only weakly correlated (r(2)=0.570, n=42, P<0.001). Cultivation efficiencies obtained with the MicroDrop technique were systematically lower than MPN values by a factor of 2.7, indicating a significant overestimation of culturability by the latter method. The composition of the cultured bacterial fraction was determined by denaturing gradient gel electrophoresis fingerprinting of 16S rDNA fragments and sequencing. This demonstrated that phylogenetically similar bacteria were recovered by both cultivation techniques. Both methods resulted in the recovery of many previously unknown aquatic bacteria affiliated to the same taxonomic groups and, in one case, in the isolation of a numerically dominant, but not-yet-cultured beta-Proteobacterium which was ubiquitous in all three lakes.