STUDY QUESTION What are the incidence and factors associated with uterine perforation by modern copper intrauterine device (Cu-IUD) and the levonorgestrel-releasing intrauterine system (LNG-IUS)? SUMMARY ANSWER Perforation incidence was similar to that reported in prior studies and did not vary between Cu-IUD and LNG-IUS groups. Lactation, amenorrhoea and a post-partum period of <6 months were common. WHAT IS KNOWN AND WHAT THIS PAPER ADDS The study supports findings in prior studies. The incidence rate was low and factors associated with uterine perforation were similar to those in earlier reports. DESIGN AND DATA COLLECTION METHOD This retrospective population-based registry study included 68 patients surgically treated for uterine perforation by an intrauterine device (IUD)/intrauterine system (IUS) at clinics in the Helsinki and Uusimaa hospital district. PARTICIPANTS AND SETTING Records of 108 patients with probable uterine perforation by an IUD/IUS were analysed, leaving 68 patients treated for uterine perforation. RECRUITMENT/SAMPLING STRATEGY Patients with diagnostic and surgical treatment codes indicating uterine perforation by an IUD/IUS between 1996 and 2009 were retrospectively selected from the Finnish National Hospital Register. DATA ANALYSIS METHOD Patients with Cu-IUDs (n = 17) and the LNG-IUS (n = 51) were analysed as one group and also compared using Mann–Whitney and chi-square tests. IUD/IUS sales numbers were used to calculate incidences. MAIN FINDINGS The overall incidence of perforation was 0.4/1000 sold devices, varying annually from 0 to 1.2/1000. The proportion of both sold and perforating LNG-IUSs increased during the study period, but perforation incidence was not affected. Demographic characteristics in the Cu-IUD and LNG-IUS groups were similar. More than half of the devices (55%) were inserted at <6 months post-partum. Breastfeeding at the time of insertion was common, comprising 32% of all patients. Moreover, of the breastfeeding women, 90% had delivered within 6 month prior to insertion. IMPLICATIONS The population-based study setting represents a good overview of patients experiencing uterine perforation with an IUD/IUS. As previously reported, the post-partum period, lactation and amenorrhoea may increase the risk of perforation. BIAS, LIMITATIONS AND GENERALIZABILITY As the study setting revealed only symptomatic patients or those attending regular follow-up, the true incidence might be somewhat higher. As there is no specific diagnostic code for uterine perforation or treatment, it is unlikely that all cases of uterine perforation can be identified in a retrospective study. STUDY FUNDING/POTENTIAL COMPETING INTERESTS Helsinki University Central Hospital research funds are acknowledged.