Malate synthase (EC 4.1.3.2) (MS), an enzyme unique to the glyoxylate cycle, was studied in cotyledons of dark-grown cotton (Gossypium hirsutum, L.) seedlings. MS has generally been regarded as a peripheral membrane protein in glyoxysomes and believed by some to be synthesized on rough ER. Immunocytochemical localization of MS in both in situ and isolated cottonseed glyoxysomes, however, showed that MS was located throughout the matrix of glyoxysomes, not specifically associated with their membranes. Biochemical data also supported matrix localization. Isolated glyoxysomes were diluted in variously-buffered salt solutions (200 millimolar KCI or 100 millimolar K-phosphate) or detergents (0.1% Triton X-100, 10 millimolar deoxycholate, or 1.0% Triton X-114) and centrifuged to pellet membranes. Greater than 70% of the MS was recovered in supematants after treatment with salt solutions, whereas generally less than 30% was released following detergent treatments. MS in pellets derived from glyoxysomes burst in low ionic strength buffer solutions was aggregated (observed on rate-zonal gradients). MS released following salt treatments was the 20S nonaggregated form indicating that salt solutions either disaggregated (or prevented aggregation of) glyoxysomal MS rather than releasing it from membranes. We confirmed reports by others that MS comigrated with ER (NADH: cytochrome c reductase) in sucrose (20-40% w/w) gradients buffered with 100 millimolar Tricine (pH 7.5) after 3 hours centrifugation. However, cottonseed MS did not comigrate with ER in gradients buffered with 10 millimolar Hepes (pH 7.0) or 20 millimolar K-phosphate (pH 7.2) after 3 hours centrifugation, or after 22 hours centrifugation in Tricine or Hepes. Collectively, our data with cotton seeds indicate that MS is not a peripheral membrane protein, and that the aggregation behavior of MS (in various buffers) very likely has led to misinterpretations of its putative associations with ER and glyoxysomal membranes.