Infection of dry beans (Phaseolus vulgaris L.) by rust, caused by Uromyces appendiculatus Pers.:Pers. (Unger), results in regular yield and quality losses in South Africa. The growing of resistant varieties is generally considered the most cost effective control measure. To meet this objective, rust resistance has been an important selection criterion since the dry bean breeding programme of the Agriculture Research Council (ARC) was initiated at Potchefstroom in the 1970's. The breakdown of resistance in earlier varieties indicated that pathogenic variation occurred in U. appendiculatus in South Africa and that a more coordinated resistance breeding programme, including race surveys, was needed. This article describes the approach that was followed at the ARC and reviews progress made with germplasm improvement by means of backcrossing. Rust resistance has been transferred from 16 donor lines, including the genes Ur-3+, Ur-5, Ur-11 and uncharacterized genes, to adapted germplasm. Results from field testing of rust infected and rust-free lines showed that resistance was considerably improved without yield loss. Incorporating the Ur-3+ gene in Teebus increased yield ca. 2.5 fold. To protect the durability of the resistance sources used, emphasis has been placed on concurrent research to combine several genes in a single well-adapted background. Attention is also being paid to race-non-specific resistance.