The treatment of acute pain with a sufficient dosage by intranasal administration of fentanyl results in a time to onset of action comparable to intravenous administration and a significantly faster onset of action than nasal titration of fentanyl. The nasal administration of a sufficient amount of fentanyl to obtain pain relief has lower maximum plasma concentrations comparable to intravenous administration and results in lower rates of adverse events like respiratory depression, nausea and vomiting. Compositions for use in the method are also disclosed.