In the study of fragmented human remains the plausibility of a cannibalism hypothesis rests primarily on the correct identification of the cause of breakage. Here the use of fracture morphologies and fragmentation indices for distinguishing green from postdepositional bone breakage is assessed using three assemblages of human bone broken by unique and well known causes, i.e., marrow fracturing of green bone, sediment pressure and impact on subfossil bone. Of the tested attributes, five appear to have diagnostic value at the statistical, assemblage level: fracture outline, fracture angle, shaft circumference, shaft fragmentation and breadth/length ratios of shaft splinters. © 1991 Academic Press Limited.