For the understanding of the mechanism of a superconducting transition, it is of great interest to clarify what the energy scale in determining the critical temperature T_c is. According to the BCS mean-field theory, the energy gap at T<<T_c, 2Δ_o, is proportional to T_c; 2Δ_o = 4.3k_BT_c for d-wave superconductors. In the case of high-T_c cuprates, however, 2Δ_o increases monotonically with the lowering of the hole-doping level p even in the underdoped region, where T_c is largely suppressed, and does not scale with T_c, except in a highly overdoped region. Interestingly, in a wide p range spreading over the under- and overdoped regions, 2Δ_o is almost proportional to crossover temperature T*, around which a spin gap and/or a (small) pseudogap of almost the same energy scale as 2Δ_o start to develop progressively; 2Δ_o/k_BT* is nearly independent of p and comparable to the BCS value in the entire p range examined. On the other hand, 2Δ_o/k_BT_c is inversely proportional to p, 2Δ_o/k_BT_c~1/p, except in the highly overdoped region, where T_c is very close to the BCS expectation. This indicates that the energy scale in determining T_c is of order pΔ_o in high-T_c cuprates. In this article, we will discuss some scenarios for the transition from pseudogap to superconducting states, where the energy scale in determining T_c is expected to be of order pΔ_o.