Background: The use of mortality and morbidity index and the length of stay (LOS) in intensive care unit (ICU) are the two primary indexes that describe the efficiency of an ICU.As the analysis of these two indexes might help in improving the efficiency of ICU, this study is undertaken. Materials and methods: An observational cross-sectional study was done on 874 patients who were admitted to ICU of Vali-e-Asr Hospital in Arak from 2002-2004. Age, gender, months of admission place of living, type of diagnosis was considered as independent variables. Mortality rate and length of stay LOS in ICU were considered as dependent variables. For analyzing the data analytic statistics such as Poisson models and mixture Poisson distribution were used.Results: From 874 patients 246 patients died, therefore mortality rate had been 28.1% in ICU. Patient's age was 40±21.8 years and LOS in ICU Was 9.2±11.9 days. The maximum and minimum mortality rate was in December and January respectively. There was a statistical significant difference between the mortality rate in this study and other studies done in other countries. Also, the median of LOS in ICU was 5 days that this value statistical is different than other studies (p<0.01). For LOS values two components Poisson mixture modeling were appropriate model. By using mixture Poisson regression method, patient's age, gender, place of living, type of diagnosis, the ward which patients transfer to ICU from it and month of hospitalization were considered as significant variables for exploring this model. The simple Poisson model was represented fewer variables for explaining LOS in ICU.Conclusion: As the mortality rate in ICU of Vali Asr Hospital is higher than universal statistics, we recommend advanced researches to solve this problem.