Objectives: As an occupational injury, percutaneous injury (PI) can result in chronic morbidity and death for healthcare workers (HCWs). A pilot surveillance system for PIs using the Chinese version of Exposure Prevention Information Network (EPINet) was introduced in Taiwan in 2003. We compared data from EPINet and recall of PIs using a cross-sectional survey for rates to establish the reliability of the new system. Methods: HCWs from hospitals that had implemented EPINet for ≥12 months completed a survey for recall of contaminated PIs sustained between October 2004 and September 2005, type of item involved, and reasons for reporting or not reporting the PI. Comparative data from EPINet for the same period were analyzed. Results: The EPINet rate, 36.1/1,000 HCW (95%CI 31.8-41.1) was almost 5 times lower (p<0.0001) than the PI recall rate for 2,464 HCWs of 170/1,000 HCWs (95%CI 155.4-185.5). Approximately 2.5 PIs were recalled for every 1,000 bed-days of care. The recall rate by physicians was 268.3/1,000, 188.5/1,000 for nurses, 88.9/1,000 for medical technologists and 81.3/1,000 for support staff. Hollow-bore needle items most commonly recorded on EPINet includ, disposable needles and syringes were underreported by 81%, vacuum tube holder/needles by 67%, and arterial blood gas needles by 75%. Nearly 63% of the reasons for underreporting were related to the complexity of the reporting process, while 37% were associated with incorrect knowledge about the risks associated with PIs. Conclusions: EPINet data underestimates a commonplace occupational injury with nearly four in five PIs not reported. Addressing the real barriers to reporting must begin with hospital administrators impressing on HCWs that reporting is essential for designing appropriate safety interventions. (J Occup Health 2009; 51: 132-136)