A model was developed for faster than real-time liquid-metal fast breeder reactor core transient analysis for purposes of continuous on-line data validation, plant state verification, and fault identification. The basic feature of this model is the use of a nodal approximation for the coolant, cladding, and fuel temperatures that gives adequately accurate power and temperature predictions with very few axial nodes. In applications of this methodology to fast loss-of-flow and overpower transients, computation times of about one-thirtieth of the real transient time per thermal-hydraulic channel were obtained. The predicted coolant and cladding temperature distributions were practically identical to those resulting from detailed finite difference computations. The predicted fuel temperatures differed by ∼1% or less from those obtained from the same finite difference computations. The analysis of the Transient Reactor Test Facility experiment TS-1C and the Experimental Breeder Reactor II experiment SHRT-17 showed very good agreement between model predictions and measurements.