In Kir6.2, mutation of three positively charged residues, R50, K185 and R201, impair ATP's ability to close the channel. The mutations do not change the channel Po in the absence of ATP, supporting involvement of these residues in ATP binding. We recently proposed that at least two of these positively charged residues, K185 and R201 interact with ATP phosphate groups to cause channel closure: the phosphate group of ATP interacts with K185 to initiate closure, while the phosphate interacts with R201 to stabilize the channel's closed state. In the present study we have replaced the three positive residues with residues of different charge, size and hydropathy. For K185 and R201, we find that charge, more than any other property, controls ATP's interaction with Kir6.2. At these positions, substitution with another positive residue had minor effects on ATP sensitivity. In contrast, substitution of K185 with a negative residue (K185D/E) decreased ATP sensitivity much more than neutral substitutions, suggesting that an electrostatic interaction between the phosphate group of ATP and K185 destabilizes the open state of the channel. At R201, substitution with a negative charge (R201E) had multiple effects, decreasing ATP sensitivity and preventing full channel closure at high concentrations. In contrast, R50E mutation had a modest effect on ATP sensitivity, and only residues such as proline and glycine that affect protein structure caused major decreases in ATP sensitivity at the R50 position. Based on these results and the recently published structure of Kir3.1 cytoplasmic domain, we propose a scheme where binding of the phosphate group of ATP to K185 induces a motion of the surrounding region, which destabilizes the open state, favoring closure of the M2 gate. Binding of the phosphate group of ATP to R201 then stabilizes the closed state. R50 on the N terminus controls ATP binding by facilitating interaction of the phosphate group of ATP with K185 to destabilize the open state.