Introduction The notification of interpersonal and self-inflicted violence (ISV) is a component of the epidemiological surveillance of this public health issue. Brazil has a large network of Health Information Systems that allows to describe the epidemiological reality and to manage the health services, but it is fundamental that the information provided by these systems be reliable and real. Thus, completeness is one of the attributes used to evaluate the quality of the information in public health surveillance systems.

Objective Evaluate the data quality of the Interpersonal and Self-inflicted Violence Notification Information System (ISVNIS) in Brazil, from 2011 to 2014.

Methods a descriptive study of the ISVNIS data was carried out considering de percentage of incompleteness of essential e mandatory variables among all variables of ISV notification form. Percentages of incompleteness were calculated on the basis of the not filled fields (filled in with ignored or blank) for each variable by year of notification. The annual average of incompleteness was calculated and classified as follow: Excellent (incompleteness <5%), Good (5%–9.9%), Regular (10%–19.9%), Bad (20%–49.9%) and Too Bad (≥50%).

Findings from 2011 to 2014, the number of ISV notifications increased 84.3%. In general, the average incompleteness of the variables analyzed was classified as Bad (22.4%), ranging from Regular for the individual notification (14.4%) and the victim identification (14.6%) variables, to Bad for closure of notification (30.8%) and case referral (32.9%) variables.

Conclusion and policy implications The ISV notification data presented bad completion quality, despite the increase in the number of notifications in the analyzed period. Data monitoring and training of professionals in the correct filling of ISV notifications is recommended.