Conformational transitions in biomolecules, especially proteins, play an important role in signaling and regulation of various biological processes. Here we propose a fast and simple method for constructing a transition pathway between two stable conformations of a protein. The protein is represented as a simplified coarse-grained (CG) model with a single site located at the C-alpha atom of each residue. The energy function of the two-state CG model is approximated by the anisotropic network model (ANM) harmonic energy surfaces of the end-states. The simple two-state energy surface comprises two local minima centered on the positions of the stable states and the system resides in one of these minima. There is a cusp with discontinuous first derivative in the multi-dimensional configuration space of the system, which acts as the transition state surface for the two-state potential. Given this simple prescription, two structures constrained to remain similar to one another but each residing on the opposite sides of the cusp, are optimized using energy minimization. This virtual “dumbbell” of two structures is then used as a crude approximation to the transition state. The pathway in the multi-dimensional space is constructed by letting these two structures slide down on their ANM surface using steepest descent energy minimization until the stable end-points are reached. This simple two-state ANM model was applied to explore the multi-dimensional and collective character of the conformational transition pathways in several systems of biological significance, including adenylate kinase, leucine transporter, sarcoplasmic reticulum Ca-ATPase and the glutamate transporter.