Purpose: The purpose of this paper is to measure efficiency of superannuation funds using data envelopment analysis (DEA), using data related to financial performance of superannuation funds. The sample comprises 183 superannuation funds covering approximately 79 per cent of the 231 largest Australian Prudential Regulation Authority (APRA)-regulated funds in 2012. The research covers a period of seven years from 2005 to 2012. The results indicate that most Australian superannuation funds are inefficient relative to the benchmark efficiency frontier based on efficient funds. The findings emphasise the importance of improving the efficiency of Australian superannuation funds by reducing overall fund expenses to narrow the gap in performance between efficient and inefficient funds. Design/methodology/approach: This study aims to contribute to policy, theory and practice in several dimensions. Member protection and the efficiency of the superannuation system are topical issues (Donald, 2009). Despite its importance from a regulatory point of view, efficiency has only been discussed in relation to operational issues such as managing agency relationships, fees and charges, investment return or economies of scale. The relative efficiency of the Australian superannuation system from an economic productivity perspective has rarely been examined, except for a study by Njie (2006), where the Malmquist productivity DEA technique was used to measure the efficiency of Australia’s retirement income system. Findings: Most inefficient funds had very low efficiency scores and were fell into the lower quintiles such as Quintiles 4 (scored 0.200-0.399) and 5 (scored 0.001-0.199). Consequently, input reduction targets were significantly higher for these two quintiles. Similarly, input reduction targets were high under the period DEA estimates. In order to be comparatively efficient, Quintile 4 funds were required to reduce total expenses by 75 per cent (−0.754) and volatility of return by 80 per cent (−0.801). Similarly, Quintile 5 funds needed to reduce total expenses by, on average, 83 per cent (−0.824) and volatility of return by 89 per cent (−0.894). Research limitations/implications: As in other empirical research, this study also depended heavily on the data collected from the secondary sources such as APRA database and other financial reports. The issues of measurement errors in data sources such as APRA database are well documented (see, e.g. Cummins, 2012). This issue needs the attention of future research on the efficiency of superannuation funds. Practical implications: The findings on individual year DEA estimates indicate that most funds were inefficient due to high expenses. Therefore, mandatory disclosure of fees and charges in a comparable manner may be necessary to justify fee payments and to address transparency and accountability issues, which are critical issues identified by the Cooper Review and the academic literature (Australian Government, 2014; Cooper et al., 2010; Gallery and Gallery, 2006). Social implications: The issue of Australian superannuation funds concentrating the majority of fund assets in highly volatile investment vehicles such as the share markets has been in the spotlight in the aftermath of the global financial crisis. There have been proposals to better diversify superannuation assets in other asset classes (Cooper et al., 2010). Originality/value: This study contributes to the current literature on superannuation funds by investigating efficiency. As efficiency studies using DEA have not been conducted on the Australian superannuation industry, this study also contributes to the academic literature on DEA and its extensive applications to various economic sectors. Efficiency scores using DEA, ranking, trends and shifts in the efficiency frontiers could be obtained for Australian superannuation funds on an on-going or annual basis.