The crucial thing to look at are the equations of motion and what they are about.
In classical electrodynamics, the equations of motion are the Maxwell equations. They can be essentially expressed by certain linear operators (differential operators) acting on the fields [itex]\vec E(\vec r,t)[/itex] and [itex]\vec B(\vec r,t)[/itex], so they are linear.
In wave mechanics, we have the Schrödinger equation where the wavefunction [itex]\psi(\vec r_1,\vec r_2,...,t)[/itex] is also acted upon by a linear operator (the Hamiltonian). Different physical systems correspond to different linear operators.
In classical mechanics, the differential operators are essentially fixed by Newton's second axiom. Instead, different physical systems correspond to different functional dependencies on the positions [itex]\vec r_i[/itex] in the equations of motion. (Here is an example of a nonlinear equation of motion:
the pendulum. Also, typical forces between interacting bodies lead to nonlinear equations of motion)
So in order to emphasize the difference between classical mechanics and wave mechanics, imagine that we have two particles and add a potential term of the form [tex]\frac{1}{|\vec r_1 - \vec r_2|}[/tex] In classical mechanics, this causes the equations of motion to become nonlinear in the [itex]\vec r_i[/itex]. In wave mechanics, the change occurs only in the linear operator which acts on the wave function. We never get something like [itex]\frac{1}{\psi}[/itex].
(Ironically, I just found out that the classical Kepler problem can be expressed in
linear form. Of course, this doesn't change the main point.)