Two blind schemes for weighted energy detection of ultra-wideband signals are proposed. The first yields the estimates of transmitted symbols by using the generalised likelihood ratio test (GLRT) paradigm, which is referred to as GLRT-WED. The second scheme first estimates the weighting vector of the WED detector by means of singular value decomposition (SVD), and then uses the obtained vector for data detection, which is referred to as SVD-WED. Both GLRT-WED and SVD-WED can achieve a performance very near to that of the ideal WED, while SVD-WED is significantly less complex than GLRT-WED.