The prediction of the critical heat flux (CHF) is an important consideration in the design and safety analysis of the sodium-cooled liquid metal fast breeder reactor (LMFBR). Many CHF correlations have been developed for light water reactor core applications. Compared with water, liquid metals show a divergent picture of boiling pattern. Therefore water CHF correlations cannot be applied directly to liquid metals. In this paper a mechanistic CHF model for liquid metals is developed based on the flow excursion mechanism. From the Baroczy correlation and the Ledinegg instability criterion a fundamental approach is tried to derive the relationship between CHF and the principal parameters. The overall mean accuracy ratio of the present model for 139 liquid metal CHF data points is 0.955, with a standard deviation of 0.155. Assessment shows that the predictions agree well with liquid metal CHF data within +/- 25% error bounds.