1. We have studied the metabolism of Ins(1,3,4,5)P4 (inositol 1,3,4,5-tetrakisphosphate) by rat liver homogenates incubated in a medium resembling intracellular ionic strength and pH. 2. Ins(1,3,4,5)P4 was dephosphorylated to a single inositol trisphosphate product, Ins(1,3,4)P3 (inositol 1,3,4-trisphosphate), the identity of which was confirmed by periodate degradation, followed by reduction and dephosphorylation to yield altritol. 3. The major InsP2 (inositol bisphosphate) product was inositol 3,4-bisphosphate [Shears, Storey, Morris, Cubitt, Parry, Michell & Kirk (1987) Biochem. J. 242, 393-402]. Small quantities of a second InsP2 product was also detected in some experiments, but its isomeric configuration was not identified. 4. The Ins(1,3,4,5)P4 5-phosphatase activity was primarily associated with plasma membranes. 5. ATP (5 mM) decreased the membrane-associated Ins(1,4,5)P3 5-phosphatase and Ins(1,3,4,5)P4 5-phosphatase activities by 40-50%. This inhibition was imitated by AMP, adenosine 5'-[beta gamma-imido]triphosphate, adenosine 5'-[gamma-thio]triphosphate or PPi, but not by adenosine or Pi. A decrease in [ATP] from 7 to 3 mM halved the inhibition of Ins(1,3,4,5)P4 5-phosphatase activity, but the extent of inhibition was not further decreased unless [ATP] less than 0.1 mM. 6. Ins(1,3,4,5)P4 5-phosphatase was insensitive to 50 mM-Li+, but was inhibited by 5 mM-2,3-bisphosphoglycerate. 7. The Ins(1,3,4,5)P4 5-phosphatase activity was unchanged by cyclic AMP, GTP, guanosine 5'-[beta gamma-imido]triphosphate or guanosine 5'-[gamma-thio]triphosphate, or by increasing [Ca2+] from 0.1 to 1 microM. 8. Ins(1,3,4)P3 was phosphorylated in an ATP-dependent manner to an isomer of InsP4 that was partially separable on h.p.l.c. from Ins(1,3,4,5)P4. The novel InsP4 appears to be Ins(1,3,4,6)P4. Its metabolic fate and function are not known.