Abstract: o characterize the tradeoff between sensing performance and energy consumption, this paper studies the robust sensing energy efficiency optimization for Reconfigurable Intelligent Surface (RIS)-assisted integrated sensing and communication systems under imperfect Channel State Information (CSI). Considering two traditional imperfect CSI models, a bounded error model and a moment-based uncertainty model, a robust optimization framework is proposed. Under the bounded error model, the worst-case sensing constraints are reformulated into linear matrix inequalities by exploiting the S-procedure. For the statistical moment-based error model, conditional value-at-risk theory is introduced to provide a convex approximation of the probabilistic sensing constraints. Moreover, the non-convex fractional EE objective is efficiently handled via the Dinkelbach algorithm, while a block coordinate descent framework is adopted to alternately optimize the base station beamforming and the RIS phase-shift matrix to maximize the sensing EE of the system. The simulation results demonstrate that the proposed method can effectively improve the sensing energy efficiency and enhance robustness compared with the benchmark scheme under different CSI error levels, transmit power budgets, and system configurations.