Human erythrocyte transketolase could be resolved from thiamin diphosphate (TDP) by acidification of the ammonium sulfate precipitate to pH 3.5, but not by other tested procedures. Resolution was 98% by chemical measurement of residual thiamin and 95% by residual enzyme activity. Reconstitution of the resolved preparation by incubation with TDP was dependent upon TDP concentration, duration, temperature, and the presence of dithiothreitol. At low TDP concentrations, 1 h was required for maximum activation; kinetic analysis then yielded an apparent Km value for TDP of 65 nM (SD 14 nM) from 100 erythrocyte lysates and similar values for reconstituted resolved preparations previously purified 400-fold and 10,000-fold. Velocity data obtained by transketolase assays in which the TDP was added to resolved preparations simultaneously with substrates yielded an apparent Km value for TDP of 2.3 microM (SD 1.6 microM) from 114 erythrocyte lysates and similar values for purified preparations. The recovery of activity following resolution and reconstitution ranged from 21 to 60% from lysates and 38 to 70% from purified preparations. Residual ammonium sulfate up to 4.9 mM decreased the apparent Km value for TDP, while a concentration of 11.3 mM increased the value in a manner competitive with TDP and with an apparent Ki value of 2.3 mM. The spectrophotometric assay of transketolase activity was greatly affected by storage of frozen solutions of the substrate ribose 5-phosphate.