Coaches believe world-ranking lists are a reliable tool for predicting international swimming performance. To examine the relationship between world-ranking and Olympic performance, we modelled world-ranking time and best time from the 2000 Olympic Games for 407 top-50 world-ranked swimmers. Analysis of log-transformed times yielded within-athlete and between-athlete coefficients of variation (CV) and percent changes in performance from world-rankings to Olympics. Variations and performance progressions were compared across sex, stroke, distance, nation and medal status. The within-athlete coefficient of variation of performance for all swimmers was 0.8% (95% confidence limits: 0.7 to 0.9%). Females were slightly less consistent, although not substantially different to males (ratio of female/male within-athlete CV: 1.1; 95% confidence limits: 1.0 to 1.2) and had a wider range of talent (ratio of female/male between-athlete CV: 1.2; 95% confidence limits: 1.1 to 1.4). Swimmers from Australia (AUS) were more consistent than those from the United States (USA) and other nations (OTHER) (ratio of within-athlete CV, USA/AUS: 1.5; 95% confidence limits: 1.0 to 2.2; OTHER/ AUS: 1.6; 95% confidence limits: 1.2 to 2.1). Most Olympic medallists (87%) had a top-10 world-ranking. Overall performance time at the Olympics was slower than world-ranking time by 0.3% (95% confidence limits: 0.2 to 0.4%), medallists improved by 0.6% (95% confidence limits: 0.4 to 0.9%) and non-medallists swam 0.6% slower (95% confidence limits: 0.5 to 0.7%). We conclude that a top-10 ranked swimmer who can improve performance time by 0.6%, equivalent to 0.13 s in the men's 50-m freestyle, will substantially increase their chance of an Olympic medal (the difference between first and fourth place).