The incidence of severe acute pancreatitis is about 30 cases per 100,000 inhabitants, and it carries an overall mortality rate of 10-15%. Infection of pancreatic necrosis occurs in 20-30% of patients with severe acute pancreatitis and triples the mortality rate. Therefore, early prediction and diagnosis of infection in necrotizing pancreatitis are extremely important. The aim of the studies included in this review was to investigate the potential of specific prognostic factors to predict the development of secondary pancreatic infection in severe acute pancreatitis. This is seen as an important tool allowing to perform a computed tomography- or ultrasound-guided fine needle aspiration for bacteriological sampling at the right moment, to confirm the diagnosis, and, finally, to select the subgroup of patients who would benefit from the antibiotic prophylaxis. Precise patients' selection could possibly result in more rational use of antibiotics in patients with acute necrotizing pancreatitis and reduction of multi-resistant bacteria. Recent studies show that C-reactive protein is an important prognostic marker of pancreatic necrosis with the highest sensitivity and negative prognostic value in this respect. Procalcitonin alone or in combination with interleukin-6 best identifies patients not at risk for infection. However, a review of the clinical studies suggests that we still do not have an optimal model, thus there is a need for new more reliable biochemical and/or clinical predictive systems.