Environmental concerns and the depletion of conventional energy sources have led to the widespread adoption of Renewable Distributed Generation (RDG) over the last decade. However, because of the intermittent nature of renewable resources, expanding RDG to a capacity level commensurate with system needs remains a major challenge for utility companies, especially in long-term planning. Within the Smart Grid (SG) paradigm, the growing diversity of demand-side resources allows the system to operate more flexibly, but it also introduces additional dynamics and uncertainty that affect the reliability contribution of RDG. This paper proposes a composite reliability framework that jointly models Renewable Distributed Generation and Demand Response (DR) to compute the Effective Load-Carrying Capability (ELCC) of RDG in a Distributed Generation System (DGS). DR is represented at two interdependent levels — an instantaneous response captured by a Gaussian Mixture Model (GMM), and a mid-term load-recovery process captured by a fuzzy model driven by a Brownian-motion discount rate — so that both the participation uncertainty and the inter-temporal rebound behavior of consumers are reproduced. An adaptive variable-step Sequential Monte Carlo Simulation (SMCS) is then used to locate, for each scenario, the reference generation capacity that reproduces the same Loss of Load Expectation (LOLE) as the RDG unit, from which the ELCC is obtained. Case studies on the IEEE 38-bus distribution test system show that, for a 50%-PV/50%-WT penetration mix, the baseline ELCC without DR is 29.51%, DR raises this value to as much as 40.28% (1.36 times higher) when DR capacity is increased by 30%, and the ELCC can reach 47.84% when the WT–DR and PV–DR correlation coefficients are 0.27 and 0.57, respectively. These results confirm that neglecting DR in ELCC studies produces a systematic and non-negligible error in the estimated reliability contribution of RDG.