Our study evaluated the differences between early and late hemodialysis (HD) initiation in the intensive care unit (ICU) according to the RIFLE (Risk, Injury, Failure, Loss, and End-stage renal failure) and AKIN (Acute Kidney Injury Network) classifications. On the assumption that early initiation of HD in critical patients according to the RIFLE and AKIN criteria decreases mortality, we retrospectively evaluated the medical records of 68 patients in our medical ICU and divided the patients into 2 groups: those undergoing HD in no risk, risk, or injury stage according to RIFLE and in stage 0, I, or II according to AKIN were defined as early HD and those in failure stage according to RIFLE and in stage III according to AKIN were defined as late HD. The median age of the patients was 66.5 years, and 56.5% were male. HD was started in 25% and 39.7% of the patients in the early stage in the RIFLE and AKIN classification, respectively. According to RIFLE, HD was started in 61.5% of the surviving patients in the early stage; this rate was 16.4% in the deceased patients (P=0.001). HD was commenced in 69.2% of the surviving patients in AKIN stages 0, I, and II and in 32.7% of the deceased patients (P=0.026). Sepsis (61.5% vs. 94.5%; P=0.001) and mechanical ventilation (30.8% vs. 87.3%; P<0.001) during HD increased ICU mortality, whereas HD initiation in the early stages according to RIFLE decreased ICU mortality (61.5% vs. 16.4%; P=0.001). In conclusion, in critically ill patients, HD initiation in the early stages according to the RIFLE classification decreased our ICU mortality.