The effects of weather conditions on children's physical activity have not been well described. PURPOSE: To evaluate the effects of meteorological variables on the number of pedometer steps accumulated by children. METHODS: Between August and December 2004 (winter to summer), 1115 Auckland children (536 boys, 579 girls; aged 5-12 yr) from 27 socioeconomically and ethnically diverse schools wore sealed multiday memory pedometers for five consecutive days (three weekdays and two weekend days). Values of daily (7 a.m. to 7 p.m.) mean ambient temperature, mean wind speed, precipitation, and duration of bright sunshine were obtained from local meteorological stations. The independent effects of each of these variables on step counts were estimated using composite mixed linear models. Effects were standardized for interpretation of magnitudes. RESULTS: Weekday and weekend-day step counts for boys were 16,100 +/- 5000 and 12,900+/- 5900 (mean +/- SD), whereas those for girls were 14,200 +/- 4200 and 11,300 +/- 4800. A 10 degrees C rise in mean ambient temperature was associated with a small increase in weekday steps [1700; 90% confidence intervals (CI) +/-1300] and a moderate increase in weekend-day steps (3400; 90% CI +/-1500) for boys, whereas for girls the effects were small (2300; 90% CI +/-1000) and unclear (-300; 90% CI +/-1200), respectively. There were substantial decreases in weekday and weekend-day steps during moderate rainfall (1.1-4.9 mm) for both sexes. Most effects of day length, wind speed, and hours of bright sunshine on step counts were trivial or unclear. CONCLUSIONS: Ambient temperature and rainfall have substantial effects on children's daily step counts and should therefore be considered when comparing physical activity across different locations or periods. Strategies to increase activity on cold or rainy days may also be appropriate.