For many polymers an effective glass transitionTg′can be defined via a change in slope of log&sgr;[&OHgr;−1cm−1] vsT−1[°K−1]. Measurements for dry cellulose acetate (CA), previously doped by soaking (48 h) in 0.1Malkali‐chloride solutions, indicate thatTg′depends on ion and polymer properties. A model based on fluctuation theory and ``free volume'' concepts is developed. The main assumptions are: (1) ions partially fill void space reducing the available local free volume; and (2) reduction in local free volume about an ion can be accounted for in the statistical expression⟨(V−V0)2⟩av=RTV0/Bby subtracting an effective ion volumeVifrom the total volume (per mole)Vof a reference aggregation of particles of most probable volumeV0. (Bis the bulk modulus,Rthe gas constant.) With simplified distribution functions, we obtained the expressionsTg≈B&dgr;2/6RV0andTg′=Tg−2a(&dgr;/m)m2Wi+am2Wi2, wherea = B/6V0R, Wiis ionic volume (per mole) from x‐ray measurements on crystals,m = Vi/Wi, and &dgr; is a measure of free volume. For CA, the best fit isTg′≈Tg−0.3Wi+0.14Wi2corresponding to a fractional free volumefg≈&dgr;/V0, which may be as much as 12%. BelowTg′, activation energies for conduction by alkali‐chloride‐doped CA were found to be given byEb[kcal / mole]≈ 15+0.7Wi≈16.1+2.45(1024&agr;), where &agr;[cm3/ion] is the electronic polarizability of the ion. AboveTg′, the energyEawas approximately 29 kcal/mole, independent of the ion. Alternative interpretations in terms of internal pressure and polarization effects are discussed.