Introduction: Over the last decades, survival of extremely preterm infants improved but there´s still significant morbidity among this group. We pretend to evaluate if specific attitudes/characteristics are associated with higher survival or survival without severe disabilities and elaborate predicting outcome models. Material and Methods: Observational descriptive study, including the 205 liveborn/stillborn infants -gestational age 22w0d-26w6d- born at an Obstetrics Unit or transferred to a Neonatology Unit of a Level III Hospital, from January-2000 to December-2009. We collected variables related to management in the prenatal/neonatal period, neonate performances and psychomotor development(18-24 months). Significant associations between variables/outcomes were identified by chi-square test or t-test; multivariate logistic regression models were used to describe and predict mortality/morbidity. Results: Advanced Gestational Age (GA) (p=0.001), antenatal corticotherapy(p=0.001), cesarean section(p=0.001), inborn delivery(p=0.021) and increased weight(p=0.001) were associated with survival. Absence of Intraventricular Hemorrhage (IVH) grade 3-4(p=0.001) and absence of Periventricular Leukomalacia (PVL) (p=0.005) were associated with survival without severe neurossensorial deficit. According to multivariable models, advanced GA (OR=0.353,CI95% 0.208-0.599), increased weight (OR=0.996,CI95% 0.993-0.999) and antenatal corticotherapy (OR=0.150,CI95% 0.044-0.510) were associated with lower mortality risk. Rupture of membranes less than 12 h duration was associated with higher mortality risk (OR=3.88,CI95% 1.406-10.680). IVH grades 3-4 was associated with higher morbidity risk (OR=16.931,CI95% 2.744-104.452). Mortality and severe morbidity models predicted correctly the outcome in 78.1% and 85.7% of the cases, respectively. Conclusions: Mortality/morbidity models might be valuable tools providing insight in the prediction of the outcome of these neonates and helping parental counseling.