A series of 44 samples of bone wedges of human origin, intended for allograft openwedge osteotomy and obtained without particular precautions during hip arthroplasty were re-examined. After viral inactivity chemical treatment, lyophilisation and radio-sterilisation ( intended to produce optimal health safety), the compressive strength, independent of age, sex and the height of the sample ( or angle of cut), proved to be too widely dispersed [10-158 MPa] in the first study. We propose a method for selecting samples which takes into account their geometry ( width, length, thicknesses, cortical surface area). Statistical methods ( Principal Components Analysis PCA, Hierarchical Cluster Analysis, Multilinear regression) allowed final selection of 29 samples having a mean compressive strength sigma(max) = 103 MPa +/- 26 and with variation [61-158 MPa]. These results are equivalent or greater than average materials currently used in openwedge osteotomy.