The K⁺ channel selectivity filter is found in the pore (P-) region of the channel and contains a consensus sequence of amino acids, TxxTxGYG. Studies in voltage gated (Taglialatela et al., 1993) and inward rectifier (Passmore 2003) channels have shown the residue preceding the GYG triplet of the signature sequence is important in determining the relative permeability of the channel to Rb⁺. In TASK-3 the equivalent residue is an isoleucine (I) in both pore domains. We have previously reported that mutation of this residue in TASK-1 alters channel selectivity (Yuill et al 2004). We have substituted a serine residue for the isoleucine at position 94 in P1 of TASK-3 and expressed the channel in Chinese hamster ovary (CHO) cells. Whole cell patch clamp was used to measure pH sensitivity, and channel selectivity was measured from the shift in reversal potential when 70mM K⁺ was replaced by Rb⁺ or Na⁺. The I94S mutant had a significant effect on the selectivity of the channel. The permeability ratio for Rb⁺/K⁺ was increased from 0.69 ± 0.004 (n=3; mean ± s.e.m.) in wild type channels, to 1.08 ± 0.02 (n=3; P<0.01) in I94S. The effect on the permeability of I94S to Na+ was striking, as wild type channels are relatively Na+ impermeant with a P_Na/P_K of 0.01 ± 0.004 (n=3). In I94S the P_Na/P_K was significantly increased to 0.97 ± 0.007 (n=3; P<0.01). The I94S mutant also displayed an altered pH dependence compared to wild type channels, which have a pK_a of 6.45 at –100mV (70mM [K+]_O). I94S reduced pH sensitivity, with a basal level of activity present between pH 5 and 6.5, indicating incomplete channel closure. The residual pH sensitivity gave a pK_a of 6.86 at 100mV (n=3). These results show that I94 is crucial for selectivity in TASK-3. They also suggest that H98 is not the only residue involved pH sensing. Therefore, like Lopes et al (2001), we also believe that the response to acidification in these channels must involve a gating process.