This paper proposes a novel database merging approach and re-examines the fundamental questions regarding hedge fund performance. Before drawing conclusions about fund performance, we form an aggregate database by exploiting all available information across and within seven commercial databases so that the widest possible data coverage is obtained and the effect of data biases is mitigated. Average performance is significantly lower but more persistent when these conclusions are inferred from the aggregate database than from some of the individual commercial databases. Although hedge funds deliver performance persistence, the average fund does not deliver significant risk-adjusted net-of-fee returns while the gross-of-fee returns remain significantly positive. Consistent with previous literature, we find a significant association between fund characteristics related to share restrictions as well as compensation structure and risk-adjusted returns.