We prove that Best Fit bin packing has linear waste on the discrete distribution Ufj kg (where items are drawn uniformly from the set f1=k 2=3 j=kg) for sufficiently large k when j = k and 0:66 < 2=3. Our results extend to continuous skewed distributions, where items are drawn uniformly on 0 a], for 0:66 a < 2=3. This implies that the expected asymptotic performance ratio of Best Fit is strictly greater than 1 for these distributions.