Using self-consistent field theory (SCFT) and Monte-Carlo simulations, we study the structure and dynamics of loops and bridges in the lamellar phase of symmetric ABA triblock copolymers at χN = 80. The bridge fraction, ν B , linearly correlates with the average variance, X 1 2 = ⟨X ̂1i 2 ⟩, of the first Rouse mode. Using SCFT with constraint X 1 2 , we calculate the free-energy landscape, F(L,X 1 2), and observe a nonmonotonic variation of the optimal lamellar spacing, L*, with X 1 2 . SCFT also provides information about the distribution, P(X ̂1i ), of Rouse modes of individual chains. In the lamellar phase, P exhibits two pronounced peaks, corresponding to the single-chain loop and bridge states. This suggests that the system can be conceived as a mixture of noninteracting loops and bridges with a two-state Markov dynamics, yielding a dynamic equation for the relaxation of the nonconserved, collective order parameter, X 1 2 . These findings are corroborated by multichain simulations of a soft, coarse-grained model. We observe an extremely long relaxation time, 4 × 10 5 τ R , compared to the Rouse time, τ R , in the disordered state. This timescale is the inverse of the conversion rates from loops to bridges and vice versa, which we obtain by single-chain simulations in conjunction with forward-flux sampling. These results suggest that the multichain simulations can be significantly accelerated by the heterogeneous multiscale method (HMM).