Frequency of weekly aerobic activity was compared with annual illness-related absenteeism in 79,070 adult US adult workers. Weekly exercise, days per week of aerobic activity (> or = 20 minutes), and absenteeism consisting of days per year and grouped as 1 to 3, 4 to 6, and 7+ days were recorded. After controlling for confounding variables, chi-squared values and odds ratios were calculated. A significant (chi 2 = 280.37) relationship was found between absenteeism and exercise. Differences (P < 0.05) in absenteeism were found between no exercise and all frequencies of weekly exercise. One day of exercise was associated with lower absenteeism when compared with no exercise, and 2 days of exercise was more favorable than one. No differences were found between any other combinations (2 to 3, 2 to 4+, 3 to 4+ days) of exercise frequency and absenteeism. These data suggest a significant relationship between exercise frequency and illness-related absenteeism.