Raltitrexed (Tomudex; TOM) hepatotoxicity is usually characterized by a transient and self-limiting increase in transaminase levels. How this may condition daily clinical practice is still unclear. The aim of this study was to investigate predictive factors of TOM hepatotoxicity. In total, 130 patients were treated at two medical oncology institutions with TOM (3 mg/m2) (52 patients) or TOM plus oxaliplatin (TOMOX) (100 mg/m2 day 1 or 70 mg/m2 day 1, 8) (78 patients). A multinomial logistic regression (adjusted for multilevel data) was performed (on all administered chemotherapy courses) to assess the dependence of hepatic toxicity on a set of clinical factors correlated with patient, disease and treatment characteristics. Creatinine clearance was calculated by the Cockcroft formula before each chemotherapy course. Most of the patients presented colorectal cancer (95%) and metastatic disease (93%). Out of the 130 patients, 41 were aged 70 or more, while 119 (91.5%) had a good performance status (PS) (ECOG 0 or 1). Before chemotherapy, liver metastases were present in 78 (60%) patients and elevated transaminase in 25 (19%). A total of 584 courses were administered (252 TOM and 332 TOMOX). National Cancer Institute Common Toxicity Criteria grade 1/2 and 3/4 transaminase toxicity was observed in 62 and 20% of patients, respectively. To control transaminase increase, glutathione (GSH) or ademethionine (SAMe) was administered in 96 and 129 cycles, respectively. Hepatotoxicity conditioned delays (a week or more) in 60 (10%) chemotherapy cycles and was the reason for the discontinuation of chemotherapy in eight (6%) patients. Among the factors evaluated with multivariate analysis, sex, age, PS, creatinine clearance, previous chemotherapy treatment, presence of liver metastases and oncology centre were not significantly associated with TOM hepatotoxicity. Elevated baseline transaminase levels (p=0.001), number of chemotherapy cycles (p<0.001), TOM cumulative dose (p=0.018), unprolonged intervals between courses (p<0.001) and TOMOX regimen (p<0.001) emerged as factors predictive of hepatotoxicity. In the same analysis, GSH (p<0.001) and SAMe (p<0.001) were hepatoprotective agents. This study confirmed TOM-based hepatotoxicity as a clinical relevant side-effect and a major factor for treatment delays or discontinuation. Predictive and protective factors listed above could assist the management of this toxicity that has probably been underestimated until now.