Soybean rust (Phakopsora pachyrhizi) is the disease of greatest concern to soybean growers in Brazil. The fungus has found the right conditions for overwintering and spreading every season throughout all soybean areas of the country. The national 'anti-rust' consortium has been responsible for the official first report of soybean rust at a given municipality each season. Using publicly available data of the reports for the southernmost states, Paraná (PR) and Rio Grande do Sul (RS) spanning 14 growing seasons (2004/2005 to 2017/2018) we fit models for describing the disease outbreak onset time and spatiotemporal dynamics. The dataset was composed of 3,596 reports in commercial fields located in 258 and 275 locations in PR and RS, respectively. The Kaplan-Meier estimates of the seasonal mean outbreak onset time (relative to sowing date) ranged from 56.6 to 118.4 days (89.3 days, all seasons). The extended K-function analysis showed a stronger evidence of space-time interaction for the first reports in RS than in PR state. The mean nearest-neighbor distances varied across seasons within a state and were most typically greater early in the season, declining afterwards, which suggests that outbreaks initiated at multiple locations and spread locally. The final size of epidemic area, calculated by connecting polygon extreme points, ranged from 50 to 200 ?103 km2 in RS and 50 to 150 ?103 km2 in PR. Further analysis including the effect of covariate on the time to outbreak and epidemic front will be presented.