Delay Tolerant Networks (DTNs) operate without immediate source-to-destination connections. Instead, nodes utilize a store-carry-forward method for traffic routing. However, flooding the network with unlimited message copies may prove ineffective under resource constraints. In contrast, quota-based approaches offer resource efficiency but may suffer from low delivery rates and significant delivery delays. This paper proposes the Adaptive Neuro-Fuzzy Message Replication Technique (ANFMRT), which dynamically adjusts message replicas based on nodes’ ability. This decision relies on node capacity factors such as encounter history, energy levels, buffer size, and time-to-live values. ANFMRT is applied to Spray and Wait, EBR (Encounter-Based Routing), and DBRP (Destination-Based Routing Protocol) protocols, optimizing message dissemination according to node capability. Simulation results show that integrating ANFMRT improves performance: delivery probability increases by 10%, while overhead ratio and average latency decrease by 26% and 10%, respectively. Furthermore, composite efficiency metrics (DO, DL, DLO) improve by 72%, 28%, and 90%.