In long-term follow-up studies of survival after an initial event (eg, an operation) mortality from causes other than the one under study obscures the results, especially in elderly patients. In the traditional approach to the calculation of expected mortality a fictitious cohort is drawn from the general population, being matched for age, sex, and calendar time at the time of the initial event. The membership of this cohort is then kept constant from the initial event until the closing date of the study. The survival and mortality of this static cohort is then compared with that of the dynamic patient cohort to throw light on mortality from extraneous causes. This method can lead to severe bias if there is a strong correlation between the duration of observation of the patients and their age. The analysis can be improved by applying rate adjustment when calculating the background component of mortality. In this approach mortality rates from the general population are adjusted (weighted) so that the age, sex, and calendar year are at all times identical with those of each of the patients still alive and under observation. This is illustrated by means of a simplified example and a real-life one from a study at survival after aortic valve replacement. Estimation of rate-adjusted background mortality provides a framework that may put long-term survival, especially of elderly patients, in proper perspective.