PURPOSE: Problem solving magnetic resonance imaging (MRI) is used to exclude malignancy in women with equivocal findings on conventional imaging. However, recommendations on its use for women recalled after screening are lacking. This study evaluates the impact of problem solving MRI on diagnostic workup among women recalled from the Dutch screening program, as well as time trends and inter-hospital variation in its use. METHODS: Women who were recalled at screening mammography in the South of the Netherlands (2008-2017) were included. Two-year follow-up data were collected. Diagnostic-workup and accuracy of problem solving MRI were evaluated and time trends and inter-hospital variation in its use were examined. RESULTS: In the study period 16,175 women were recalled, of whom 906 underwent problem solving MRI. Almost half of the women (45.4%) who underwent problem solving MRI were referred back to the screening program without further workup. The sensitivity, specificity, and positive and negative predictive values of problem solving MRI were 98.2%, 70.0%, 31.1%, and 99.6%, respectively. The percentage of recalled women receiving problem solving MRI fluctuated over time (4.7%-7.2%) and significantly varied among hospitals (2.2%-7.0%). CONCLUSION: The use of problem solving MRI may exclude malignancy in recalled women. The use of problem solving MRI varied over time and among hospitals, which indicates the need for guidelines on problem solving MRI.