We describe a device for random settling preparation of calcareous nannoplankton (coccolith) samples. The device allows easy draining and cleaning, as well as sedimentation at different heights of the water column through the use of different slide carriers. Reproducibility and accuracy of the device has been tested with standardised microbeads as spiking material. In applying counting techniques with different equations for determining absolute abundances, we discovered major anomalies in the calculated results that we interpreted as being due to the effect of convection currents within the device and the usage of elevated cover slides. A modified formula that corrects for the influence of elevated cover slides in random settling experiments is therefore proposed. Although the settling method is more time-consuming than standard smear-slide techniques, additional information is gained about the spatial and temporal distribution of coccoliths. These are important for palaeoecological and palaeobiogeographical interpretations.