Doses of radiolabeled water, salicylic acid, lidocaine and diazepam were applied in 2 ml of phosphate buffer saline to the exposed dermis on the thigh of a perfused rat hindlimb. Hindlimbs were perfused with Krebs-Heinseleit buffer at 37 degrees C containing 4% albumin or 2.5% dextran 40 at a constant rate of 4 ml/min. Clearance of solutes from dermal sites was monitored by frequent sampling at the dermal site and of the outflowing perfusate. In addition, the concentration of solutes in the various tissues below the treated and contralateral sites were determined at the end of a perfusion after dissection, solubilization and scintillation counting of the individual tissues. Estimates of the absorption rate constant from the dermis and the elimination rate constant from the perfused limb were made by the simultaneous fitting of absorption and efflux data with a one-compartmental pharmacokinetic model. Diazepam was cleared fastest through the dermis, followed by water, salicylic acid and lidocaine. The efflux of diazepam and salicylic acid from the hindlimb was significantly enhanced by the presence of albumin in the perfusate relative to no albumin (P < .05), whereas the efflux of water and lidocaine were unaffected by changes in albumin content. Analysis of individual tissues at the end of the perfusion showed that water had the highest relative tissue concentrations with diazepam the least, high concentrations appearing to partition into and remain associated with the dermis. The presence of albumin in the perfusate significantly increased the concentration of diazepam found in contralateral skin samples (P < .05) and showed the same trend for salicylic acid and lidocaine. These findings show that protein binding significantly affects the deep tissue penetration and distribution of dermally applied solutes and that this effect is more prominent for highly protein-bound solutes.