Background: FOLFIRI regimen, which is composed of 5-FU, Leucovorin, and Irinotecan, is used in the first-line chemotherapy of metastatic colorectal cancer. Irinotecan life threatening toxicity is partly related to cytotoxic drug metabolite which is primarily inactivated by the UGT1A1 enzyme. The primary aim of the present research was to find the correlation between UGT1A1-genotype and clinical toxicity of irinotecan. Methods: In a prospective study from March 2011 to December 2013, all patients with metastatic colorectal cancer who had been referred to Medical Oncology Department of Iran Cancer Institute were genotyped for UGT1A1*28 before the first cycle of chemotherapy. All of the patients signed informed consent and trial approved by Ethics Committee of the Tehran University of Medical Sciences. Reduction of the standard dose of Irinotecan (180 mg/m2 body surface area) was measured based on NCI toxicity criteria after the first cycle of chemotherapy. Patients with previous treatment with Oxaliplatin and fluorouracil (5-FU) in the adjuvant setting and adequate liver, kidney, and heart function were included in the trial. Both synchronous and metachronous metastatic disease were noticeable. Results: A total of 50 patients with median age of 52 years were included. Most (70%) of the patients had more than one site of metastases in peritoneum, liver, and/or lung. Thirty-one patients had UGT1A1*1 normal genotype, 13 were in heterozygote and 6 were in homozygote state ofUGT1A1*28/*28. A clinically relevant increase in early toxicity was found in patients carrying the UGT1A1*28/*28 genotype with odds Ratio (OR) of 2.6 (95%CI 2.5-27.28). Similarly, there was a trend of lower overall survival in homozygote group with an HR (Hazardous Ratio) of 2.76 (95%CI .88-.61). No statistically significant relationship was found between UGT1A1genotypes and response to therapy. Conclusions: UGT1A1 28*/28* is strongly associated with drug’s life-threatening toxicity even in a moderate dose of Irinotecan. On the other hand, UGT1A1 genotype data was not helpful to differentiate response to treatment.