A simple procedure for assay of Ins(1,4,5)P3 5-phosphatase is described. The reaction products [( 3H]Ins(1,4)P2, [3H]InsP and myo-[3H]inositol) are completely separated from one another, with quantitative yield, on Amprep SAX (100 mg) minicolumns. [3H]Ins(1,4,5)P3 [and [3H]Ins(1,3,4,5)P4] are adsorbed to the columns but not released to any appreciable extent by the elution conditions used. In GH3 cells, the stepwise dephosphorylation of [3H]Ins(1,4,5)P3 to myo-[3H]inositol was demonstrated, and was inhibited by 2.3-bisphosphoglycerate. The Km of the soluble form of the enzyme was lower in GH3 cells (8-13 microM) than in IMR-32 cells (26-32 microM) or in rat cerebral-cortical samples (22 microM. The Km of the particulate form of the enzyme was similar in all three preparations (10-16 microM). The pH profiles of the two soluble 5-phosphatases differed, with a wider pH optimum for the GH3-cell activity than for the IMR-32-cell activity. The soluble and particulate GH3 enzymes were more sensitive than the corresponding IMR-32 enzymes to inhibition by p-hydroxymercuribenzoate, whereas there were no differences in their sensitivities to glucose 6-phosphate, 2,3-bisphosphoglycerate, fructose 1.6- and 2.6-bisphosphate and non-radioactive Ins(1,3,4,5)P4. Dialysis of the soluble fractions and washing of the particulate fractions did not affect the inhibitor sensitivities, except for the soluble IMR-32 fraction and p-hydroxymercuribenzoate. The Km value of the soluble GH3 5-phosphatase activity was lower, and the inhibition by Ins(1,3,4,5)P4 greater, after adsorption to and elution from phosphocellulose. It is concluded that there are qualitative differences in the properties of the soluble 5-phosphatase activity from GH3 and IMR-32 cells.