ABSTRACT. To increase knowledge of particle dynamics in indoor environments, we have conducted experiments on the effects of small surface discontinuities and roughness on deposition from natural convection flow. Measurements were made in a half-height (1.22 m) aluminum test chamber and in a full-scale experimental room. In the test chamber, air flow was induced by uniformly heating the floor and one wall while cooling the ceiling and opposite wall to a constant temperature difference of 3 K. In the full-scale room, one wall was heated and the opposite wall was cooled to a constant wall-to-wall temperature difference of 3 or 7 K. Other surfaces in both experiments were approximately adiabatic. Near-monodispersed fluorescent particles (diameters 0.1, 0.5, or 1.3 p m in the half-height experiments and 0.2 or 1.0 p m in the full-scale experiments) were injected into the chamber. Following an exposure period, the mass of fluorescent particles deposited on sections of the walls and/or plates mounted on the walls were extracted and measured by fluorometry. The effect of surface discontinuities was explored by comparing deposition onto the walls or onto flush-mounted plates with deposition onto thin, smooth, surface-mounted plates. The effect of surface roughness was investigated by measuring deposition onto textured plates (finely scratched, a rectangular array of 2.4 mm balls, or skip-coat drywall texture). Deposition of the smallest particles (0.1 and 0.2 pm) was relatively insensitive to surface obstructions and texture, but the effect of roughness increased with particle size. For 1.3 p m particles, deposition to the roughest surface was as much as five times greater than deposition to a smooth wall. The effect of surface roughness was greater for vertical surfaces than horizontal and for warm surfaces than cool. Deposition velocities measured with a 3 K temperature difference are fairly consistent between the full-scale room and the half-height chamber. Overall, surface roughness of the type commonly found indoors can significantly impact deposition rates, and, therefore, many real surfaces cannot be assumed to be smooth when analyzing particle deposition in indoor environments.