It is not feasible to use in vivo tetracycline double labeling to study bone formation in biopsies taken during the emergency fixation of fractures. We therefore compared the trabecular localization and extent of osteoblastic alkaline phosphatase (AP) perimeters with tetracycline and osteoid perimeters in iliac crest biopsies from 7 women with postmenopausal osteoporosis and 13 women without metabolic bone disease. Fresh biopsies were chilled to -70 degrees C, and triplicate serial unfixed undecalcified cryostat sections were cut and reacted for AP, stained for osteoid, or mounted unstained. At individual remodeling sites, the mineralizing perimeter (M.Pm) was measured as the extent of a double or single label accompanied by greater than or equal to 1 lamella of osteoid and greater than or equal to 1 lamella of mineralized matrix between the mineralization front and the adjacent label. Osteoid perimeters (O.Pm) and AP perimeters (AP.Pm) were also measured. In each biopsy there was good agreement between the location of AP and bone formation (kappa statistic, range 0.71-1.0). The overall sensitivity and specificity of AP as an indicator of the location of bone formation were 0.963 and 0.902, respectively. At the level of the basic multicellular unit, in those samples in which greater than 3 active BMUs were found, there was (1) significant positive correlation between the M.Pm and both AP.Pm and AP-positive O.Pm (except 1 patient) and (2) no significant difference between the M.Pm and AP-positive O.Pm (17 of 18 patients and 18 of 18 patients at the tissue level).(ABSTRACT TRUNCATED AT 250 WORDS)