We have augmented our previous studies [Storey, Shears, Kirk & Michell (1984) Nature (London) 312, 374-376] on the subcellular location and properties of Ins(1,4,5)P3 (inositol 1,4,5-trisphosphate) phosphatases in rat liver and human erythrocytes. We also investigate Ins(1,3,4)P3 (inositol 1,3,4-trisphosphate) metabolism by rat liver. Membrane-bound and cytosolic Ins(1,4,5)P3 phosphatases both attack the 5-phosphate. The membrane-bound enzyme is located on the inner face of the plasma membrane, and there is little or no activity associated with Golgi apparatus. Cytosolic Ins(1,4,5)P3 5-phosphatase (Mr 77,000) was separated by gel filtration from Ins(1,4)P2 (inositol 1,4-bisphosphate) and inositol 1-phosphate phosphatases (Mr 54,000). Ins(1,4,5)P3 5-phosphatase activity in hepatocytes was unaffected by treatment of the cells with insulin, vasopressin, glucagon or dibutyryl cyclic AMP. Ins(1,4,5)P3 5-phosphatase activity in cell homogenates was unaffected by changes in [Ca2+] from 0.1 to 2 microM. After centrifugation of a liver homogenate at 100,000 g, Ins(1,3,4)P3 phosphatase activity was largely confined to the supernatant. The sum of the activities in the supernatant and the pellet exceeded that in the original homogenate. When these fractions were recombined, Ins(1,3,4)P3 phosphatase activity was restored to that observed in unfractionated homogenate. Ins(1,3,4)P3 was produced from Ins(1,3,4,5)P4 (inositol 1,3,4,5-tetrakisphosphate) and was metabolized to a novel InsP2 that was the 3,4-isomer. Ins(1,3,4)P3 phosphatase activity was not changed by 50 mM-Li+ or 0.07 mM-Ins(1,4)P2 alone, but when added together these agents inhibited Ins(1,3,4)P3 metabolism. In Li+-treated and vasopressin-stimulated hepatocytes, Ins(1,4)P2 may reach concentrations sufficient to inhibit Ins(1,3,4)P3 metabolism, with little effect on Ins(1,4,5)P3 hydrolysis.