To evaluate the feasibility of quantitatively assessing regional skeletal fluoride uptake in humans in focal and generalized bone disease, we investigated the skeletal kinetics of [18F]fluoride ion with dynamic PET imaging. Dynamic image sets were acquired over a 60-min interval in a multiplane PET device, and input functions (plasma 18F time-activity curves) were measured directly from arterialized blood and, in some cases, determined from image-derived left ventricular cavity activity measurements. Our results indicate: 1. A steady-state ratio of [18F]fluoride ion concentration in plasma to whole blood greater than unity (1.23 for plasma to directly assayed whole blood and 1.44 for plasma to left ventricular cavity imaged concentrations. This concentration difference produces a scaling factor that must be considered when using image derived or directly measured input functions. 2. The preferred tracer kinetic model configuration for [18F]fluoride ion skeletal kinetics is a three compartment model that includes a "bound" and "unbound" bone [18F]fluoride ion compartment. 3. The rate constant for forward transport of [18F]fluoride ion from plasma to the extravascular space of bone (K1) and the regional blood volume parameter generate estimates of bone blood flow and vascular volume, respectively, that are in the physiologic range of reported for mammals. Estimates of the uptake constant for fluoride in bone, using nonlinear regression (KNLR = 0.0360 +/- 0.0064 ml/min/ml), are in very good agreement with an estimate of the same parameter obtained with Patlak graphical analysis (KPAT = 0.0355 +/- 0.0061 ml/min/ml). 4. Generating parametric images of KPAT facilitates quantification of regional bone [18F]fluoride ion kinetics. The method is computationally practical, and, with either the parametric imaging approach or with standard region of interest analysis, can be used to generate quantitative estimates of fluoride uptake (a "bone metabolic index") in focal skeletal regions or in more generalized distributions.