On the basis of the present author's original theory on dissociative negative surface ionization (DNSI), the degree of dissociation, γ(ZX), of incident sample molecule (ZX) and that of negative ionization ϵ − (X), of atom or radical (X) to be converted to negative ion (X −) on a heated surface are evaluated as a function of surface temperature, T, or work function, φ, of the surface in order to predict the best experimental condition under which both γ(ZX) and ϵ − (X) are essentially unity and under which the electron emission current accompanied with DNSI is negligibly small (less than 5%) compared with the emission current of X −. When the incident beam flux, N(Cl 2), or the partial pressure, P(Cl 2), of Cl 2 adjacent to the ionizing surface (to which the sticking probability, σ(Cl 2), of Cl 2 is taken as 10 −3) is 10 13 molecules cm −2 s −1 or ∼ 4 × 10 −8 torr, for example, the best condition is satisfied by selecting φ and T in the ranges of ∼ 2–3 eV and ∼ 500–800 K, respectively. In the case of HCl on a surface of φ ⋍ 3 eV, for example, selection of T ⋍ 1060, 1200 and 1400 K is concluded for σ(HCl) N(HCl) = 10 10, 10 12 and 10 14 molecules cm −2 s −1 [or P(HCl) ⋍ 3 × 10 −9, 3 × 10 −7 and 3 × 10 −5 torr for σ(HCl) ⋍ 10 −2], respectively. Alkali earth metal oxides (SrO, BaO) or LaB 6 instead of refractory metals (W, Ta, Nb) are recommended to be used as ionizing surface material because the former has a much smaller φ and Richardson constant than the latter. As a practical method for selecting the best condition for DNSI, the present theory is applicable to any sample molecule (ZX) for a given value of N(ZX) or P(ZX) so long as the following data are available: (1) both entropy and enthalpy changes due to dissociation of ZX and to negative ionization of X and (2) σ(ZX) for the surface of interest.