The spread of grey leaf spot caused by Cercospora zeae-maydis from infested maize residue on the soil surface and progress over time were studied under tropical conditions at two locations in central Uganda, Kabanyolo and Namulonge, using a susceptible local cultivar, LP16. Infested maize residue collected the previous season was used as the inoculum source in varying amounts; 80% coverage, 40% coverage and 0% soil coverage were used to simulate no tillage, minimum tillage and maximum tillage with complete burial of residues. In all the seasons, disease spread was adequately described by a power law model, whereas disease progress over time was equally well represented by both Gompertz and logistic models. The time factor was highly significant (P < 0.05) for disease development, indicating both increases in disease intensity over time and change in gradients with time. The amount of infested residue on the soil surface in the maize planting was highly significant (P < 0.05) for slope (b) and intercept (a) of the linearised disease gradient. The gradient parameter (b) based on the average for the four directions from the residue source ranged from -0.96 to -0.08 in the second cropping season of 1999, from -2.16 to -0.01 and from -0.87 to -0.03 in the first and second seasons of 2000, respectively. The distance from the infested residue significantly affected foliar disease, but the direction from the residue inoculum source did not, nor was the distance by direction interaction significant. However, interactions between amount of residue and distance were significant. Percentage leaf area affected by grey leaf spot, the areas under disease progress curves and disease intercepts decreased with distance at both locations, but rate of disease increase (r) was generally constant with distance from residue source at Kabanyolo, but not at Namulonge. The relationship between disease severity and amount of residue cover appeared to be affected by the level of background inoculum. At Kabanyolo, where little background inoculum was present, differences were detected in disease severity parameters between the no-residue plots and residue plots, whereas at Namulonge, where high background levels existed, there was no effect of residue on disease.