Microdialysis catheters (CMA-60 with a polyamide dialysis membrane; 20,000-molecular wt cutoff) were either immersed in an external medium or were inserted in the quadriceps femoris muscle of healthy subjects, using perfusate with or without dextran 70. Varying the position of the outflow tubing induced changes in hydrostatic pressure. The sample volumes were significantly smaller in catheters perfused without a colloid compared with those perfused with a colloid [11-50% (in vitro) and 8-59% (in vivo) lower than in colloid-perfused catheters with the same position of the outflow tubing]. The sample volumes were also significantly smaller when the dialysis membrane was influenced by maximal hydrostatic pressure (above position) compared with minimal hydrostatic pressure (below position) [7-38% (in vitro) and 3-46% (in vivo) lower than in catheters in the below position with the same perfusion fluid]. In vivo, glucose concentration at a perfusion flow rate of 0.33 microl/min was higher when the catheters were perfused without a colloid [18-28% higher than in colloid-perfused catheters with the same position of the outflow tubing (P < 0.001)] than with a colloid. A corresponding difference also tended to occur with lactate, glycerol, and urea. At 0.16 microl/min, the glucose concentration was the same irrespective of whether fluid loss had been counteracted by colloid inclusion or by lowering of outlet tubing. The mechanism behind the observed concentration difference is thought to be a higher effective perfusion flow rate when fluid loss is prevented at low-perfusion flows. This study shows that fluid imbalances can have important implications for microdialysis results at low-perfusion flow rates.