In most countries, numbers of deaths rise considerably during the winter season. This winter excess in mortality has, however, been declining during recent decades. The causes of this decline are hardly known. This paper attempts to derive a number of hypotheses on the basis of a detailed description of trends in winter excess mortality in the Netherlands. Numbers of death by age, cause of death, and month of death for the period 1953-1988 were analysed by log-linear regression. There was a 50% reduction in winter excess mortality between the 1950s and 1970 followed by a much smaller reduction in later years. The decline in winter excess for total mortality can largely be attributed to parallel declines for a number of cardiovascular and respiratory diseases. Excess mortality decreased for winter months without influenza epidemics as well as for those with epidemics. The results suggest that the decline in winter excess mortality in the Netherlands can only partly be explained by decreases in influenza-associated mortality. It is argued that the role of the introduction of central heating is minimal and that a fundamental role is played by factors closely related to socioeconomic progress.