One hundred and ninety-five mid-shaft femoral bone samples were collected at the Victorian Institute of Forensic Pathology, Melbourne, Australia. These samples came from individuals aged 1-97 years, who had died unexpectedly and were considered healthy immediately before death. From the samples, transverse sections 100-200 <mu m thick were cut on a sawing microtome. Three groups of sections [from persons aged 19-21 (n = 9), 39-41 (n = 9), and 59-61 years (n = 9) at death] were adjusted to a thickness of 100 +/- 5 mu m by hand-lapping and microradiographs were taken of these sections. Automatic image analysis methods were used to measure the area and perimeter of the femoral cortex for all 195 sections, Additionally, the number and areas of all holes within the cortex were measured for those sections that were microradiographed. The results show that, even though the number of Haversian canals measured from the entire bone cortex changes with age, the spread of values at any given age makes these measurements of limited value for age determination. From this it may be inferred that all methods that rely, to any extent, upon measurements of this parameter are likely to be flawed.