There are currently no reliable biomarkers of ageing. A biomarker should indicate biological age, that is, the amount of an animal's total lifespan it has lived and, therefore, the amount of time it has remaining. Some potential biomarkers cannot be validated as their measurement involves harm or death of the animal, such that its ultimate lifespan cannot be determined. A non-destructive biomarker would allow us to test molecular markers potentially involved directly in the ageing process, to monitor the effectiveness of therapeutic interventions to delay ageing, and provide a useful measure of general health of the organism. In the model organism Drosophila, various behavioural phenotypes change directionally with age, but we do not know whether they predict lifespan. Here we measure activity and sleep parameters in 64 wild type male flies from two recently wild-caught populations over the course of their natural lives, and determine whether such measures may predict biological age and ultimate lifespan. Indices of sleep fragmentation and circadian rhythm were the best predictors of lifespan, though population differences were evident. However, when used to predict a biological age of 50 % lifespan elapsed our best behavioural measure was slightly less accurate and less precise compared with using chronological age as predictor.