DNA labelling provides an exciting tool for elucidating the in vivo dynamics of lymphocytes. However, the kinetics of label incorporation and loss are complex and results can depend on the method of interpretation. Here we describe two approaches to interpreting labelling data. Both seek to explain the common observation that the estimated death rate of lymphocytes is higher than their estimated proliferation rate. In the first approach, an additional source of lymphocytes is postulated. In the second, it is maintained that lymphocyte heterogeneity is sufficient to account for the observation. We explain why we favour the second approach, arguing that the addition of a large source of lymphocytes is unnecessary and difficult to reconcile with what is currently known about lymphocyte physiology. We discuss how the choice of model can affect data interpretation.