We present new observational results on high-velocity clouds (HVCs) based o
n an analysis of the Leiden/Dwingeloo H I survey. We cataloged all HVCs wit
h N-HI greater than or equal to 1 .10(9) cm(-2) and found 252 clouds that f
orm a representative flux limited sample. The detailed analysis of each ind
ividual HVC in this sample revealed a significant number of HVCs (nearly 20
%) having simultaneously a velocity and a column density gradient. These HV
Cs have a cometary appearance in the position-velocity representation and a
re called henceforward head-tail HVCs (HT HVCs). The head is the region wit
h the highest column density of the HVC, while the column density of the ta
il is in general much lower (by a factor of 2-4). The absolute majority of
the cataloged HVCs belongs to the well known HVC complexes. With exception
of the very faint HVC complex L, all HVC complexes contain HT HVCs. The HT
HVCs were analyzed statistically with respect to their physical parameters
like position, velocity(nu(LSR), nu(GSR)), and column density. We found a l
inear correlation between the fraction of HVCs having a head-tail structure
and the peak column density of the HVCs. While there is no correlation bet
ween the fraction of HT HVCs and nu(LSR), we found a dependence of the frac
tion of HT HVCs and V-GSR. There is no significant correlation between the
fraction of HT HVCs and the parameters galactic longitude and latitude. The
HT HVCs may be interpreted as HVCs that are currently interacting with the
ir ambient medium. In the context of this model the tails represent materia
l that is stripped off from the HVC core. We discuss the implications of th
is model for galactic and extragalactic HVCs.