Background: The prediction of death in intensive care units is done by using scoring systems (eg APACHE II) which assess disease severity. This study was performed in Intensive Care Units (ICUs) of Boushehr’s Fatemeh Zahra hospital to evaluate the efficacy of APACH II system and also to compare the observed mortality with the predicted mortality rate and also to that of some creditable centers. Material and Methods: This cross-sectional, descriptive and analytical study was performed on one hundred patients with critical conditions on the day of their admission to the ICU wards. Data were analysed by Chi Square, student T, sensitivity, specificity, positive and negative predictive values, ROC curve and spearman correlation coefficient using SPSS version 13. Results: The mean score in internal medicine ICU (60 patients) was15.45 and in surgery’s ICU (40 patients) was 11.1. There was a positive correlation between the acquired score and mortality (p<0.001 and correlation coefficient=0.4). Mortality in our ICUs was more than that of more developed centers with respect to APACHE II score. The observed mortality rate was 31% and the predicted death rate was 19.79%. The area under ROC curve was 0.76 (CI95%=0/65-0/86). There was also a positive correlation between the acquired score and duration of ICU admission (p=0.009 and correlation coefficient=0.262). Conclusion: The APACHE II score is appropriate for predicting mortality in our ICUs. Our observed mortality rate was greater than the predicted death rate, in comparison to some developed centers which show no significant difference. Therfore it appears that we must improve our intensive cares to reduce mortality.