To expand on Arpad's answer and offer a small correction:
It is indeed possible to excite an atom with a multiphoton process. You don't even need the intermediate state to be real, it can be 'virtual'. As long as the $n$ photons add up to the energy difference between states, it is possible to excite an electron in an atom to the higher energy state. For example: Say the energy difference between the ground and first excited state is $E=\hbar\omega = 2\pi\hbar*563THz$, then you can excite the transition with a single 532nm photon or two 1064nm photons.
You can also excite from one excited state to another, higher, excited state (like in your question). This is not necessarily 'unlikey' (..."the probability to excite the electron to an even higher energy level is little") as Arpad said; the intermediate state could be long-lived state (like a metastable state), or you can shine both lasers on your sample at the same time so that the necessary two photons arrive in quick succession.
If you take a path integral approach to the problem, the decay actually happens along all paths (in your example: A,B,C). The paths interfere constructively or destructively to give you the probability of a certain pathway. Some paths will dominate (be the most likely). You don't know which decay pathway is the most likely until you measure it over and over. With enough measurements, you will have a statistical distribution showing you the various likelihoods of different paths. In principle one could calculate this too.