The equilibrium thermodynamics calculus of cell adhesion developed by Bell et al. (1984, Biophys. J. 45, 1051-1064) has been extended to the general non-equilibrium case. In contrast to previous models which could only compute the end results of equilibrium states, the present theory is able to calculate the kinetic process of evolution of adhesion, which may or may not approach towards equilibrium. Starting from a basic constitutive hypothesis for Helmholtz free energy, equations of balance of normal forces, energy balance at the edge of the contact area and rate of entropy production are derived using an irreversible thermodynamics approach, in which the restriction imposed by the Second Law of Thermodynamics takes the place of free energy minimization used by Bell et al. (1984). An explicit expression for adhesion energy density is derived for the general transient case as the difference of the usable work transduced from chemical energy liberation from bond formation of specific crosslinking molecules and the repulsive potential of non-specific interactions. This allows the energy balance to be used as an independent boundary equation rather than a practical way of computing the adhesion energy. Jump conditions are obtained from the conservation of crosslinking molecules across the edge of adhesion region which is treated as a singular curve. The bond formation and lateral motion of the crosslinking molecules are assumed to obey a set of reaction-diffusion equations. These equations and the force balance equation within the contact area, plus the jump conditions and the energy balance equation at the edge form a well-posed moving boundary problem which determines the propagation of the adhesion boundary, the separation distance between the two cell membranes over the contact area as well as the distributions of the crosslinking molecules on the cell surfaces. The behavior of the system depends on the relative importance of virtual convection, lateral diffusion and bond formation of the crosslinking molecules at the edge of the adhesion region, according to which two types of rate limiting cases are discussed, viz, reaction-limited and diffusion-limited processes.