A new approach to bundle critical power predictions is presented. In addition to a very accurate critical heat flux (CHF) model, correction factors that account for the effects of grid spacers, heat flux nonuniformities, and cold walls, which are needed for critical power predictions for practical fuel bundles, are developed. By using the subchannel analysis code COBRA IIIC/MIT-1, local flow conditions needed as input to CHF correlations are obtained. Critical power is therefore obtained iteratively to ensure that the bundle power value from the subchannel analysis will cause CHF at only one point in the bundle. Good agreement with the experimental data is obtained. The accuracy is higher than that of the W-3 and EPRI-1 correlations for the limited data base used in this study. The effects of three types of fuel abnormalities, namely, local heat flux spikes, local flow blockages, and rod bowing, on bundle critical power are also analyzed. The local heat flux spikes and flow blockages have no significant influence on critical power. However, rod bowing phenomena have some effect, the severity of which depends on system pressure, the gap closure between adjacent rods, and the presence or absence of thimble tubes (cold walls). A correlation for the influence of various rod bowing phenomena on bundle critical power is developed. Good agreement with experimental data is shown.