We studied the timing of occurrence of 1676 sporadic, community-acquired cases of Legionnaires' disease in England and Wales between 1993 and 2008, in relation to temperature, relative humidity, rainfall, windspeed and ultraviolet light using a fixed-stratum case-crossover approach. The analysis was conducted using conditional logistic regression, with consideration of appropriate lag periods. There was evidence of an association between the risk of Legionnaires' disease and temperature with an apparently long time lag of 1–9 weeks [odds of disease at 95th vs. 75th centiles: 3·91, 95% confidence interval (CI) 2·06–7·40], and with rainfall at short time lags (of 2–10 days) (odds of disease at 75th vs. 50th centiles: 1·78, 95% CI 1·50–2·13). There was some evidence that the risk of disease in relation to high temperatures was greater at high relative humidities. A higher risk of Legionnaires' disease may be indicated by preceding periods of warmer wetter weather.