There is little reported information on loads, speeds and times used by dentists to provide a good surface finish to direct filling materials or the optimum levels for these variables. This study was in two parts, clinical and laboratory. Firstly, the optimum loads, speeds and times used by general practitioners were determined using four grades of abrasive disc on three types of filling material (amalgam, composite and glass-ionomer cement). The load and duration were determined from a universal load-testing machine to which discs of the filling material were attached. The speed of the hand-piece was set by the practitioner. In order to examine to what level of roughness, for patient comfort, restorations should be finished, a study was undertaken to determine to what degree patients were able to detect roughness with their tongue. The second part of the study determined the optimal polishing conditions in the laboratory, using a specially constructed jig that permitted the applied load to be varied in a controlled manner. The effects of load, speed and duration on the surface finish were determined using two and three-dimensional profilometry with scanning electron microscopy being used for comparison. In addition the effect of the polishing on the temperature rise within the material was investigated together with the effect on this of lubrication. It was found the practitioners finished restorations using a wide range of loads, speeds and times for all the materials. Polishing amalgam, they used a very similar loading for all grades of disc. In contrast in the laboratory the optimum load decreased with decrease in grit size, the practitioners mean load being greater than optimum with fine discs. The practitioners increased speed with reducing grit size but only at the finest grade were their speed equal to optimum. The time taken by practitioners was always close to optimum. Polishing composite, practitioners always used loads appreciably higher than optimum and faster speeds, but durations were similar to optimum ones. Glass-ionomer was polished at higher than optimum loads and at slower speeds. The times taken were shorter except with the larger grit size. The temperature developed in amalgam and composite increased as the surface roughness decreased. Polishing glass-ionomer with lubricants, other than water, produced more heat than using unlubricated discs.