In 1955, 234 men and 116 women died from bladder cancer in the Netherlands. In 1988 the numbers were 794 and 317 respectively. After adjusting for the ageing of the Dutch population since 1955, female mortality rates per 10(5) person-years appear to be very stable: 2.9 from 1955 to 1959 and 3.0 from 1985 to 1988. By contrast, an increasing trend exists in males. From 1955 to 1959 and from 1985 to 1988, bladder cancer mortality rates per 10(5) person-years were 7.5 and 12.4 respectively. For men in particular, increasing mortality (and incidence) rates are seen all over the world. In many cases this increasing trend is thought to originate from a higher risk of dying from bladder cancer in successive birth cohorts rather than from a higher risk in successive calendar periods. This so-called cohort effect is explained by changes in smoking behaviour in the male population. Statistical modelling of bladder cancer mortality data from 1955 to 1988 in the Netherlands shows that the increasing temporal trend in men can also be described as a cohort effect. The risk of dying from bladder cancer increases from the 1875 birth cohort to the 1910 birth cohort, but decreases thereafter. It is concluded that this decreasing risk for generations born after 1910 will probably result in a decreasing trend in mortality in the near future, when more and more of these "youngsters" reach the age of 70+.