The U.S. Military Services have compiled an impressive collection of lessons learned. A warfighter can study only so many lessons learned, and even fewer can be recalled at a critical decision point. This paper addresses the challenge of retrieving the right lesson learned at the right time, and how we are responding to that challenge in developing the Military Analogical Reasoning System (MARS). MARS includes a corpus of operational vignettes (i.e., battle stories) that draws on a comprehensive military ontology for representation of vignette components. The corpus can be searched using a similarity-based retrieval method—structure mapping—to find a vignette from the corpus (the base vignette) that is structurally analogous to a vignette of interest (the target vignette). Calculation of the similarity quotient involves mapping patterns of events between the target and the base. Besides identifying similar vignettes, MARS generates a lesson learned that applies to both of the vignettes. The lesson learned is abstracted from the event pattern(s) that the base and target have in common. A real-time vignette-building capability is being developed to process incoming live data, such as Blue Force Tracker (which tracks the movements of friendly forces). The incoming sensor data are assembled into vignettes-so-far. Since the vignette-so-far describes the evolving situation, the comparison of the vignette-so-far to the library of vignettes results in retrieval of lessons learned relevant to the developing situation; in other words, just-in-time lessons learned. This paper describes the development and capabilities of this research prototype, as well as reviewing related research and recommending steps for further development of a just-in-time lessons learned capability.