Changes in the weather are highly significantly correlated with changes in death rate from pneumonia but the association is not a simple one. Two apparently separate periods of weather change are negatively correlated with pneumonia. Both are centred around a time approximately a week before death but one is associated with periods of changed weather of about a week's duration and the other with long-term, probably seasonal, change. Both are almost certainly related to the onset or acquisition of the disease. In addition there is an immediate positive correlation of high humidity and temperature with deaths. This occurs at a time when the patients have had the disease for a week or more and points to the need to avoid these conditions when nursing patients with pneumonia. All three associations are more marked in the elderly and all three operate at all temperatures and humidities met with in Britain. Control of the microenvironment of the elderly is necessary throughout the temperature range.