During the past five years, we have witnessed an avalanche of research on the adverse health effects of air pollutants. After the major visible pollution of smoke emission had been dealt with, and when controls were in place for reductions in emissions from coal burning, the industrial world was tempted to assume that the problem of air pollution, which had culminated in the London episode of December 1952, no longer existed. Of course, ozone formation was recognized to be a problem in Los Angeles. But, stimulated by “timeseries” studies of the daily levels of particulate pollution and the concurrent daily deaths or hospital admissions or emergency visits, we have learned that fine particles (less than 10 microns in size or PM 10 ) are associated with increased mortality from respiratory and cardiovascular causes. They are also associated with increased hospital admissions for asthma, pneumonia, chronic obstructive lung disease, and cardiovascular conditions; with depressed lung function and increased respiratory symptoms in children; and possibly with significant changes in heart rate in the elderly. It has been difficult to keep abreast of the many studies that have been published; but the consistency of these in places with very different populations, climates, and other concomitant pollutants, has led to the opinion of many observers that a causal relationship must be assumed, even in the absence of an understanding of the biological mechanism. Canadian data from Toronto and Southern Ontario have contributed importantly to this information, and much of this has been work published by Health Canada. This issue of the Journal contains the most recent of these studies by Burnett and his colleagues (pages 152-56). They studied the daily deaths, excluding accidents and suicides, from 11 Canadian cities including all the major centres, with a total population of 9.85 million people, from 1980 to 1991. As one would expect, the pollutants measured, CO, NO 2 , SO 2 , and O 3 , were highly inter-correlated. Daily measures of particle mass, such as PM 10 , were not available for analysis and could not be included in the multiple pollutant models. The authors report that NO 2 was the pollutant most strongly associated with daily mortality, with ozone, sulphur dioxide and carbon monoxide also significantly associated but less strongly. The authors have concluded that ambient air pollution “generated from the burning of fossil fuels is a risk factor for premature mortality in Canadian cities”, and they suggest that their estimates can be used to determine public health benefits due to air pollution mitigation strategies. Their work confirms the many studies which have indicated that combustion products are currently responsible for a variety of adverse health effects. It is unfortunate that comprehensive PM 10 data were not available for this study since it might have been found that this pollutant had a stronger association than ambient NO 2 , which, being an indicator of exposure to vehicle exhaust, might well be a surrogate for it. There are a number of reasons suggesting that the observed outcome should not be directly attributed to NO 2 ; in these low concentrations it seems unlikely that it could be having a direct effect. Furthermore, there have been recent examples in which NO 2 was at far higher concentrations than are found in these Canadian cities, but in which dramatic effects on mortality did not occur. In London, England, for example, in December 1991, NO 2 reached a level of 400 ppb for eight consecutive hours on two consecutive days. There was a small blip in mortality, but it seemed likely that this could be accounted for by the concomitant rise in PM 10 . This was not being measured at the time, but as visibility was obviously reduced in the episode, levels of over 200 micrograms/m were probable. The international literature, particularly the data coming from the ambitious European project known as APHEA, does contain some evidence of associations between various health outcomes and NO 2 , but the data are not consistent across different cities. One of the difficulties is that although exposure to indoor NO 2 from gas cooking has been carefully quantified, we do not have much information on the general relationship between individual exposures and the results of ambient monitoring; what there is, from a study of schoolchildren in Denmark, suggests that outdoor monitors do reflect personal exposures reasonably faithfully. The interest of the media in the adverse health effects of air pollutants at current levels, is ephemeral. Current data indicate that the public health community cannot be complacent about this problem; air monitoring should be not only maintained, but improved; and research, using modern large databases, should be continued. In the present climate of budget cuts, it is important to preserve the research capability in this field which the group at Health Canada has exploited with such distinction.