Objectives We assessed the association between public health midwives perceived barriers to identify intimate partner violence (IPV) and their identification of IPV victims in Sri Lanka.

Methods We carried out a community-based cross-sectional study in the Kandy district, Sri Lanka between August and September 2009. Participants were all the public health midwives in the Kandy district (n=425). We collected data on midwives identification of IPV victims during the past 3 months, and their agreement to six possible barriers for the identification of IPV. We analysed data using the multivariate logistic regression analysis; we adjusted the six barriers for midwives work duration, education and marital status.

Results Of the 425 midwives, 48 (11.3%) reported that they did not identify any IPV victim during the past 3 months. Midwives were less likely to identify IPV victims if they agreed that I don't have adequate knowledge to detect or help IPV victims (adjusted OR (AOR)=0.30, 95% CI=0.12 to 0.73), or I don't feel that I can help an IPV victim (AOR=0.17, 95% CI=0.07 to 0.43). However, we did not observe such an association between the following barriers and midwives identification of IPV: my work load is too much, women are always accompanied by someone, asking on IPV might offend women and though identified, there are no supportive services for IPV victims.

Conclusions Sri Lankan midwives were less likely to identify IPV victims if they perceived that they do not have an adequate knowledge or the ability to identify and support the IPV victims.