Age- and sex-specific death rates from multiple sclerosis were calculated both from all conditions coded and from those coded as underlying or primary cause of death for the United States by color for 1955, and for Denmark (1963–68) and Norway (1956–60). Primary-cause death rates for the United States (1959–61) were also assessed. Underlying-cause deaths accounted for about 3/4 the total reported in each land as well as for each sex in Denmark and Norway, but there were marked differences in this proportion between the sexes in the United States. For total deaths with MS all three lands had essentially equal death rates by sex, though for underlying-cause MS death rates there had been a female preponderance in US whites and in Denmark. From underlying-cause data, age-specific MS death rates in the US attained almost a plateau from about age 45 on, in Denmark there was a sharp and high peak at about age 62, and Norway's rates were intermediate to the other two lands. In all enumerations the proportion of deaths with MS provided by underlying-cause deaths fell with increasing age, but otherwise the age-specific rates from total deaths with MS (all conditions coded) showed curves similar to the underlying-cause rates in Denmark, Norway, and US white females; however the US white males then showed a striking maximum from age 65–84. In each land, there was a tendency for female MS primary-cause age-specific death rates to anticipate the male by some five to ten years, which did persist to some extent when associated-condition deaths were added. That such discrepancies should be apparent between total deaths and primary-cause deaths for a disorder with such a high proportion of deaths considered as underlying adds yet another aspect to be considered when evaluating disease from mortality data. Complete enumeration of all deaths encoded to each separate rubric of the 8th revision of the International Statistical Classification (ISC) for at least one year would permit some assessment of the effects that coding as to cause of death might have on inferences as to geographic or other selective factors in disease.