The detection of nanometre-sized meteoric smoke particles in the Earth’s mesosphere and lower thermosphere is difficult. The particles are too small for optical detection and so far only the charged fraction of the particles has been probed by rocket-borne instruments. One way to obtain maximum information about the smoke particles is direct sampling with rocket-borne particle samplers. The MAGIC project (Mesospheric Aerosol – Genesis, Interaction and Composition) aims to quantitatively answer fundamental questions about the properties of smoke in the atmosphere. The first launch of such particle samplers was performed during the MAGIC rocket campaign from Esrange, Sweden, in January 2005. In order to characterise the sampling process, we have performed simulations of the trajectories of nanometre-sized dust particles towards the MAGIC detectors with a new particle motion model. An important feature of this model is the Brownian motion of the particles due to thermal collisions of the gas molecules. As a result, we obtain the detection efficiency for the MAGIC detectors as a function of altitude and particle size. Our simulations confirm that particles of radii down to 0.75 nm impact on the sampling surface with an efficiency exceeding 80% over the entire mesospheric altitude range of interest.