This paper is devoted to the phase matching and parametric gain properties of the ${\ensuremath{\chi}}^{(3)}$ mixing, and points out several specific behaviors whenever compared to ${\ensuremath{\chi}}^{(2)}$ interactions. For a given pulsation ${\ensuremath{\omega}}_{4}$ and a fixed direction of propagation, phase matching simultaneously exists for numerous sets $({\ensuremath{\omega}}_{1},{\ensuremath{\omega}}_{2},{\ensuremath{\omega}}_{3}).$ Two combinations of the interacting pulsations exist: ${\ensuremath{\omega}}_{4}={\ensuremath{\omega}}_{1}+{\ensuremath{\omega}}_{2}+{\ensuremath{\omega}}_{3}$ and ${\ensuremath{\omega}}_{4}+{\ensuremath{\omega}}_{1}={\ensuremath{\omega}}_{2}+{\ensuremath{\omega}}_{3},$ both with ${\ensuremath{\omega}}_{4}>{\ensuremath{\omega}}_{1},{\ensuremath{\omega}}_{2},{\ensuremath{\omega}}_{3}.$ The second scheme has no equivalent among the three-wave processes; in this scheme, phase matching exists for all configurations of polarizations. It is then possible to make use of the largest elements of the ${\ensuremath{\chi}}^{(3)}$ tensor. Furthermore, in noncentrosymmetric crystals, solutions exist where nonzero ${\ensuremath{\chi}}^{(3)}$ nonlinear effective coefficient is associated with a zero ${\ensuremath{\chi}}^{(2)}$ nonlinear effective coefficient. Concerning the gain of ${\ensuremath{\chi}}^{(3)}$ parametric amplification, the conversion efficiency is analytically calculated for the most general case: it is nonzero only if at least three fields are incident on the crystal. The calculated gain allows us to deduce the oscillation threshold of a ${\ensuremath{\chi}}^{(3)}$ optical parametric oscillator and the effect of the pulse duration: the threshold is not independent of the parametric fluorescence field, so that it can only be calculated analytically for an injection-seeded optical parametric oscillator.