Effective communication requires that consumers attribute the message content to its intended source. The proposed framework distinguishes four types of source identification processes-cued retrieval, memory-trace refreshment, schematic inferencing, and pure guessing-and delineates their contingencies. Two experiments examine portions of the framework, and experiment 2 introduces a new methodology for decomposing multiple processes. Findings suggest that when cued retrieval fails, consumers try to refresh the original memory trace for the learning episode-a process that is effortful. They invoke schematic inferencing only if the original memory trace cannot be refreshed. Reliance on cued retrieval seems to require little processing capacity. If there is some motivation for accuracy, pure guessing appears to be invoked only as a last resort.