In the case of SARS-CoV-2 infection, children seem to be less affected than adults, but data regarding epidemiologic characteristics and biochemical values are poor and essentially based on limited case series. The aim of our study is to highlight the predictive value of some biochemical markers at hospitalization, for the correct classification of the patient in the form of disease. Methods: We performed an analytical retrospective study on 82 pediatric patients diagnosed with COVID-19 in the emergency department, with moderate or severe form of disease, and treated in our tertiary hospital. We analyzed the epidemiologic characteristics, symptomatology and biochemical values and compare the data according to the form of disease. Results: The mean age at admission was 4.5 years (median 1 year) and the masculine/feminine ratio was 1.5. Comparing the data between the two groups of patients (42 severe/40 moderate), we observed that the severe form presented with a lower pH at admission (p = 0.02), hyperglycemia (p = 0.01), increased values of transaminases (p = 0.01 and 0.02) and hypoproteinemia (p = 0.01). Also, the severe form was statistically significantly associated with comorbidities, acute respiratory distress, rising of the inflammatory markers during hospitalization. Hyperlactatemia (Lactate > 1.5 mmol/L) was significantly associated with the age under one year (p < 0.001). Mortality rate was 9.75% and the median age at death was 3 months. Univariate logistic regression model shows that the presence of anemia increased the probability of death 88 times, comorbidities 23.3 times and ketoacidosis 16.4 times. Conclusions: Metabolic acidosis, hyperlactatemia, hyperglycemia, modified hepatic values and hypoproteinemia are biochemical markers associated with the severe form of disease in SARS-CoV-2 infection in children. Presence of anemia, comorbidities and ketoacidosis are important risk factors for death of pediatric patients with SARS-CoV-2 infection.