Tumour lysis syndrome (TLS) can be a life threatening complication during induction chemotherapy in patients with acute myeloid leukaemia (AML). However, its incidence, risk stratification and management strategies are based on data mainly from lymphoid malignancies. Risk-adapted prophylactic strategies for TLS are in need with the introduction agents like Rasburicase. This single-centre retrospective chart review study was conducted to identify TLS predictive factors in a large AML population undergoing first induction chemotherapy. Between January 1980 and December 2000, 614 consecutive adult patients were diagnosed of de novo AML and started therapy in our institution. Prophylaxis of TLS consisted of intravenous hydration and oral allopurinol. Laboratory TLS (LTLS) was defined as either a 25% change from baseline or level above upper laboratory normal values (ULN) (K+>5meq/l, Uric Acid >7,5mg/dl, Phosphate >5meq/l), or below lower laboratory normal values (Ca2+<8meq/l) for any two or more of this parameters; or creatinine >1,4mg/dl and level above ULN of at least one of the previously defined parameters. These criteria must be met within 3 days before and 7 days after the initiation of chemotherapy in absence of any other recognisable cause. Clinical TLS (CTLS) was defined as the presence of LTLS and at least one of the following complications: oliguric renal failure, hemodyalysis, signs of hyperkalemia in EKG, cardiac arrhythmia, tetania, or seizures. We used the variables identified in a multivariate analysis to construct a scoring system to predict TLS. Overall, 101 patients (17%) developed TLS. Of them 72 (12%) developed LTLS and 29 (5%) developed CTLS. LTLS did not impact induction death rate (21% vs 24%, p=0.51), but CTLS was associated with higher induction death rate (83% vs 24%, p<0.001). In 14 patients (2%) CTLS was considered a major cause of death. Univariate analysis showed that CTLS was significantly associated with FAB M4-M5, hepatomegaly, splenomegaly, GOT >50UI/l, Creatinine >1.4mg/dl, Uric acid >7.5mg/dl, WBC>25x109/l, and LDH >1xULN. In order to perform a multivariate analysis previous selected variables were stratified as follows: WBC>25 and ≤75x109/l, >75x109/l; Creatinine >1.4mg/dl; Uric acid >7.5mg/dl; LDH >1 and ≤4xULN, >4xULN. According to the hazard ratio of each variable a prognostic score system for CTLS was stated, giving 3 points for: Creatinine >1,4mg/dl, Uric acid >7,5mg/dl, WBC>75x109/l, or LDH >4xULN; and 1 point for: WBC>25 and ≤75x109/l or LDH >1 and ≤4xULN. CTLS incidence was 1%, 9% and 25% in patients from 0–3, 4–5, and ≥6 points respectively. 80% of patients were correctly classified, with a sensitivity of 83%, when an individual score of 4 was used as cutpoint. The observed area under the ROC curve was 87%. When our predictive model was introduced in a multivariate analysis, it remained as the solely independent prognostic factor. In conclusion, this study shows that TLS is a frequent alteration found in AML during induction therapy. Nevertheless only one third of patients who met laboratory criteria developed CTLS dramatically increasing death rate in induction. Using four routinary available laboratory data, we have developed a predictive model able to identify different CTLS risk patients subsets. This model could be useful to establish risk-adapted strategies for prophylaxis against TLS.