PIP: The premise of this discussion is that a systematic and continuous monitoring system is required to assemble data on the social indicator "socio-economic differences in mortality." Attention is directed to 5 particular types of data: secular trends; class differentials and age; linearity versus dichotomy; cross-cutting variables; and downward mobility and biological selection. The following 2 basic questions are examined and answered with a qualified "yes:" 1) does the health care system have any relevance to mortality differentials; and 2) can a health care system have any degree of meaninful autonomy from the overall social system. The policy implications of this analysis are reviewed in terms of the value content of medical education, the organization of the health care system, the emphasis on health, and the focus on the community. The concepts of control and power are analyzed as the key to socioeconomic differentials. Emphasis on differential exposures to "stressors" is rejected for what is termed "a sense of coherence" -- a global orientation which emerges, or fails to emerge among the lower classes, against the background of a high level of generalized resistance resources. Essentially the problem is that the constricted, emergency, powerless, and unpredictable character of lower social class existence prevents individuals of lower class and groups from being able to cope with stressors. Ways that the health care system can strengthen the sense of coherence of the lower classes include the following: a formal monitoring system in each society; caution in assuming that technological advances, environmental control, and health education are egalitarian in their consequences; and the need to identify high-risk groups within the lower classes.