Membranes of Streptococcus lactis were solubilized with 1.1% octyl-beta-D-glucopyranoside in the presence of 0.37% acetone/ether-washed phospholipid from several sources. After adding excess Escherichia coli phospholipid as bath-sonicated liposomes, phosphate:sugar phosphate antiport was reconstituted in proteoliposomes by a 25-fold dilution in 0.1 M KPi (pH 7). Assays of 32Pi:Pi exchange showed that antiport was subject to an inactivation which varied in severity according to the lipid present at solubilization. Recovery of Pi-linked exchange was improved by the presence of 10-20% glycerol or other osmolyte during extraction. The osmolytes tested in this regard have included polyols (glycerol, erythritol, xylitol, sorbitol), sugars (glucose, trehalose), and two amino acids (glycine, proline). Each gave 10--20-fold increased recoveries of 32Pi:Pi antiport compared to controls using only detergent and lipid; these precautions were not required for the efficient reconstitution of F0F1-ATPase. Antiport in the artificial system was studied most carefully when glycerol was the stabilizing additive. For that case, the Kt values for Pi or 2-deoxyglucose 6-phosphate transport (275 and 25 microM, respectively) were the same as in native membranes. Maximal rates of Pi and 2-deoxyglucose 6-phosphate transport (200 and 42 nmol/min/mg of protein, respectively) and the turnover number for Pi exchange (25--50/s) suggested that antiporters were recovered without loss of activity. We conclude that the quantitative aspects of bacterial anion exchange are amenable to study in an artificial system, and that the use of osmolytes as general stabilants can be a valuable adjunct to current techniques for reconstitution of integral membrane transport proteins.