Dimensional characteristics of lumbar vertebrae (L1-L5) were obtained by examination of three osteologic collections containing a total of 304 male and 181 female vertebrae. The multimetric (14 parameters) discriminative analysis has demonstrated that the highest accuracy of ordinal localization based on the general diagnostic model (L1-L5) was obtained for L5 (93.76%) and the lowest one for L2 (56.45%). Correction of the results using several discrimination models improved the accuracy of classification to 76.65% (L1), 81.61% (L2), 79.82% (L3), and 81.99% (L4). Diagnosis of gender identity of isolated lumbar vertebrae by multimetric discriminative analysis proved to be possible in 87.7-88.8% of the cases; this figure amounted to 93.8% if the totality of these vertebrae was included in the analysis. The accuracy of body length measurements using multiple regression equations was estimated at +/- 6.22-6.45 cm for individual vertebrae and +/- 0.62 cm based on the complete set of them. In all these cases, sex and age of the subjects should be taken into consideration. It is concluded that the proposed methods can be used in forensic medical practice for the identification of skeletonized bone remains.