Abstract Background: It has been proposed that a systematic pain assessment increases the awareness of the need to treat and prevent pain, and most international and national neonatal pain guidelines state that pain assessment should be performed in a systematic way. National surveys show a wide variation in compliance to these guidelines. Methods: A survey to all Swedish neonatal units was performed in 1993, 1998, 2003 and 2008, concerning the use of, and need for, pain assessment tools. Results: The number of units that tried to assess pain increased from 64% in 1993 to 83% in 2008. Forty-four per cent of these used a structured method in 2003, compared to three per cent in 1998. The most common pain indicator was facial actions. Conclusion: The proportion of neonatal units that reported the use of a structured pain assessment tool has increased significantly from 1993 to 2008. There is a need for better evidence for the relation between the implementation of pain guidelines and the actual performance of pain assessment.