Fault of lamps in tunnel partially reduces road illuminance. Therefore, it becomes difficult to drive comfortably and safely. Thus, it is important for road administrators to quantitatively estimate fault probability of lamps and its effect. In this paper, the final target is to evaluate the risk of the reduction of road illuminance, caused by fault of lamps. Accordingly, the statistical methodology is constructed to analyze the influence and range of road illuminance the lamps have. Concretely, the relation between lamps and road illuminance is formulized as a spatial distribution of illuminance model, which takes into account of spatial interaction, and Bayesian estimation is proposed to estimate the unknown parameter and autocorrelation parameter of a spatial distribution of illuminance model. Finally, the validity of the proposed methodology is vertified through an application study using actual inspection data of a tunnel.