Programs to increase the life expectancy of old people are becoming increasingly relevant in an aging society. Their adequacy depends on the quality of accounting death causes. Objective: To evaluate the quality of accounting of death causes for population over the working age. For 98 061 deaths of people over working age registered in Moscow dead database, the structure of underline and multiple causes of death were compared. To determine whether information on morbidity of elderly can improve the diagnosis of death causes the structure of death causes was compared with officially registered prevalence and detected prevalence. The last was calculated on the basis of the household survey of health of elderly population in Nizhny Novgorod Region (22 558 people). We find fundamental difference between the structure of causes for officially registered prevalence, detected prevalence and mortality. In the structure of death causes the nervous diseases are in the second place and the proportion of uncertain death causes is 6.7%. The difference in the structure of underline and multiple causes of death is not so great. For people over working age, the coding errors were detected in 9.8% underline death causes. Thus, morbidity statistics cannot provide informational support for diagnosing causes of death in full. To improve the quality of accounting of death causes, it is advisable to introduce the institution of coders and to give them the possibility of verifying death diagnoses in medical organizations where the diagnosis has been established.