Prompted by recent solar and atmospheric data, we re-analyze the four-neutrino description of current global neutrino oscillation data, including the LSND evidence for oscillations. The higher degree of rejection for non-active solar and atmospheric oscillation solutions implied by the SNO neutral current result as well as by the latest 1489-day Super-K atmospheric neutrino data allows us to rule out (2+2) oscillation schemes proposed to reconcile LSND with the rest of current neutrino oscillation data. Using an improved goodness of fit (gof) method especially sensitive to the combination of data sets we obtain a gof of only 1.6 times 10^{-6} for (2+2) schemes. Further, we re-evaluate the status of (3+1) oscillations using two different analyses of the LSND data sample. We find that also (3+1) schemes are strongly disfavoured by the data. Depending on the LSND analysis we obtain a gof of 5.6 times 10^{-3} or 7.6 times 10^{-5}. This leads to the conclusion that all four-neutrino descriptions of the LSND anomaly, both in (2+2) as well as (3+1) realizations, are highly disfavoured. Our analysis brings the LSND hint to a more puzzling status.