We measured the pharmacokinetics of hydromorphone in normal volunteers given three doses of the drug (10, 20, and 40-mu-g/kg as intravenous 45-s injections on different days. Concentrations of hydromorphone in plasma from serial blood samples were measured by a high-performance liquid chromatography method specific for hydromorphone with a detection limit of 0.1 ng/mL. In all cases, plasma hydromorphone concentration versus time data for individual subjects were best described by a triexponential (instead of mono- or biexponential) function. Furthermore, we found that the pharmacokinetics of hydromorphone was independent of dose across the range studied. Averaged across doses, the distribution and terminal elimination half-lives were 1.27 min (t1/2-pi), 14.7 min (t1/2-alpha), and 184 min (t1/2-beta), respectively. Average values for systemic clearance, initial dilution volume, and steady-state volume of distribution were 1.66 L/min (Cl), 24.4 L (Vc), and 295 L (Vd(ss)). Our results indicate that hydromorphone pharmacokinetic parameters are linear across a fourfold range of doses that are usually employed clinically and that previously reported pharmacokinetic values for hydromorphone (based on radioimmunoassay measurements) deserve reconsideration.