Avian erythrocytes were incubated with myo-[3H]inositol for 6-7 h and with [32P]Pi for the final 50-90 min of this period. An acid extract was prepared from the prelabelled erythrocytes, and the specific radioactivities of the gamma-phosphate of ATP and of both the myo-inositol moieties (3H, d.p.m./nmol) and the individual phosphate groups (32P, d.p.m./nmol) of [3H]Ins[32P](1,3,4,6)P4,[3H]Ins[32P](1,3,4,5)P4, [3H]Ins[32P](3,4,5,6)P4 and [3H]Ins[32P](1,3,4,5,6)P5 were determined. The results provide direct confirmation that one of the cellular InsP4 isomers is Ins(1,3,4,5)P4 which is synthesized by sequential phosphorylation of the 1,4,5 and 3 substitution sites of the myo-Ins moiety, precisely as previously deduced [Batty, Nahorski & Irvine (1985) Biochem. J. 232, 211-215; Irvine, Letcher, Heslop & Berridge (1986) Nature (London) 320, 631-634]. This is compatible with the proposed synthetic route from PtdIns via PtdIns4P, PtdIns(4,5)P2 and Ins(1,4,5)P3. The data also suggest that, in avian erythrocytes, the principle precursor of Ins(1,3,4,5,6)P5 is Ins(3,4,5,6)P4. Furthermore, if the gamma- (and/or beta-) phosphate of ATP is the precursor of the phosphate moieties of Ins(3,4,5,6)P4, then this isomer must be derived from the phosphorylation of Ins(3,4,6)P3. If the gamma- (and/or beta-) phosphate of ATP similarly acts as the ultimate precursor to all of the phosphates of Ins(1,3,4,6)P4, then, in intact avian erythrocytes, the main precursor of Ins(1,3,4,6)P4 is Ins(1,4,6)P3. This contrasts with the expectation, based on results with cell-free systems, that Ins(1,3,4,6)P4 is synthesized by the direct phosphorylation of Ins(1,3,4)P3.