This study aims at explaining loss of separation events over the Swedish air space in 2011- 2012, which occurred despite an automated conflict detection tool working as designed. The study suggests that there may be a trade-off between spending visual scan time on own conflict detection versus visual scan time spent on examining potential conflicts presented by the conflict detection automation. The issue is hard to solve, and is unfortunately far from resolved. This area of research requires a substantial applied research effort, if the goal is to both increase safety and capacity of ATM through the use of automation.