Storage oils in seeds are synthesized via the Kennedy pathway localized in the endoplasmic reticulum (Gurr, 1980). Triacylglycerol is formed by the sequential addition of three acyl groups to a glycerol 3-phosphate backbone. The esterifications are catalysed by three separate acyltransferase enzymes. It is believed that the enzyme catalysing esterification of acyl groups to the sn-2 position of glycerol, acyl-CoA: lysophosphatidic acid acyltransferase (LPA.AT ), plays an important role in regulating the acyl composition of seed oils. LPA.AT has the greatest degree of substrate specificity for acyl-CoAs of the three acyltransferases. Selectivity is a measure of substrate preference shown when the enzyme is incubated with a mixture of fatty acyl-CoAs. When incubated with a mixture of saturated and unsaturated fatty acylCoAs (Stobart & Stymne, 1985), safflower LPA.AT shows the following preference: linoleate > linolenate > oleate, with no incorporation of saturated fatty acids at position sn-2. However, if the levels of unsaturated fatty acids are limiting, the LPA.AT can use any available saturated fatty acids. It does not show absolute specificity for unsaturated fatty acids. Therefore the fatty acyl composition of lipids depends on a complex balance of selectivities and specificities. Complete exclusion of some fatty acyl-CoAs from position sn-2 has been demonstrated (Sun et al., 1988). Maturing seeds of palm, maize and rapeseed were incubated with lauroyl( 12:0), oleoyl( 18: 1 ) and erucoyl(22: 1 j CoAs. No 22:l could be incorporated at position sn-2 in any of the plants tested, even in rapeseed, which can contain up to 65% erucic acid in triacylglycerol. Also, only the palm was capable of esterifying 129 CoA to sn-2. The other two acyltransferases showed much lower substrate preferences. It is important to know the substrate preference of these acyltransferase enzymes, before strategies for modifying oil composition of seed storage oils by genetic manipulation can be worked out. If the composition of the cellular fatty acylCoA pool is modified, it is necessary to ensure that the existing acyltransferases can utilize the available fatty acids. Introducing acyltransferases with different substrate preferences from other plant species is a possible solution to this problem. Preliminary results on substrate specificity showed that the LPA.AT from crude rapeseed microsomal preparations had a four-fold preference for oleoyl-CoA over palmitoylCoA. This indicates that the acyltransferase activity in the membrane preparation was localized mainly in the endoplasmic reticulum. A second Kennedy pathway exists in the plastid, having a different set of enzymes from that of the endoplasmic reticulum. The LPA.AT activity associated with the inner membrane of the plastid has a preference for unsaturated over saturated acyl groups (Cronan & Roughan, 1987). This difference in the substrate specificities provides a means of testing that the purified protein does in fact originate in the endoplasmic reticulum, and that the plastid enzyme has not been inadvertently purified. After sucrose density gradient centrifugation, most of the LPA.AT activity was found in fractions with the same buoyant density as that of the endoplasmic reticulum. This supports the results on