a = a[~(np.isnan(a).any(axis=1))] # removes rows containing at least one nan a = a[~(np.isnan(a).all(axis=1))] # removes rows containing all nan