We discuss the possibility that galactic gravitational wave sources might give burst signals at a rate of several events per year, detectable by state-of-the-art detectors. We are stimulated by the results of the data collected by the EXPLORER and NAUTILUS bar detectors in the 2001 run, which suggest an excess of coincidences between the two detectors, when the resonant bars are orthogonal to the galactic plane. Signals due to the coalescence of galactic compact binaries fulfill the energy requirements but are problematic for lack of known candidates with the necessary merging rate. We examine the limits imposed by galactic dynamics on the mass loss of the Galaxy due to GW emission, and we use them to put constraints also on the GW radiation from exotic objects, like binaries made of primordial black holes. We discuss the possibility that the events are due to GW bursts coming repeatedly from a single or a few compact sources. We examine different possible realizations of this idea, such as accreting neutron stars, strange quark stars, and the highly magnetized neutron stars ("magnetars") introduced to explain Soft Gamma Repeaters. Various possibilities are excluded or appear very unlikely, while others at present cannot be excluded.