Two soluble and two membrane bound phosphatidylinositol (PI) specific phospholipases D (EC 3.1.4.4) were identified, partially purified and characterized from Catharanthus roseus suspension cultured cells. In order to compare the chromatographic properties of the four enzymes, column chromatography was performed under the same conditions for all four enzymes. The soluble enzymes have apparent molecular weights of 132 kD and 48 kD, the plasma membrane bound enzymes of 132 kD and 58 kD. Based on their origin and their molecular weight the types were designated S132, S48, PM132 and PM58, respectively. Isoelectric points of the native type S132, S48, PM132, PM58 phospholipases D were 4.89, 4.89, 4.6 and 4.8. After ammonium sulphate precipitation in the presence of Triton X-100, the cytosolic enzymes precipitated, whereas the plasma membrane localized enzymes floated up. In the presence of Triton X-100 all enzymes do not depend on divalent cations. Maximum activity is obtained at a Triton X-100 PI ratio of 8:1 and half maximum activity at a ratio of 14:1 to 20:1. Using Triton X-100 PI mixed micelles, the 48 and 58 kD enzymes depended on the surface concentration of PI as well as on the bulk concentration of PI, whereas the 132 kD proteins depended nearly exclusively on the surface concentration of PI. A detailed analysis of the 48 kD and 58 kD phospholipases show that they very closely follow the surface dilution kinetic model as proposed by Deems et al. (R.A. Deems, B.R. Eaton, E.A. Dennis. J. Biol. Chem., 250 (1975) 9013–9020). The interfacial Michaelis constants ( K m B) for phosphatidylinositol were 0.017 mol fraction for the 48 kD cytosolic protein and 0.062 for the 58 kD membrane-bound phospholipase, respectively. As has been reported for microsomal PI-specific phospholipases D (A. Becher, J.B. Wissing, K.G. Wagner. Plant Sci., 97 (1994) 143–151) all these enzymes do not display any transphosphatidylation activity. The formation of diphosphatidylglycerol was also not observed. Further Mg 2+ or Ca 2+ are not needed for activity, EDTA, EGTA and DTPA show no effect. The pH optimum is 5.25 for the low molecular weight and 6.5 for the high molecular weight isoforms. Half maximum activity is observed at pH 4.2 and 6.2 for the smaller enzymes and at pH 4.5 and 7.5 for the larger enzymes.