The first generation of long-baseline laser interferometric detectors of gravitational waves will start collecting data in [2001][2002][2003]. We carefully analyse their planned performance and compare it with the expected strengths of astrophysical sources. The scientific importance of the anticipated discovery of various gravitatinal wave signals and the reliability of theoretical predictions are taken into account in our analysis. We try to be conservative both in evaluating the theoretical uncertainties about a source and the prospects of its detection. After having considered many possible sources, we place our emphasis on (1) inspiraling binaries consisting of stellar mass black holes and (2) relic gravitational waves. We draw the conclusion that inspiraling binary black holes are likely to be detected first by the initial ground-based interferometers. We estimate that the initial interferometers will see 2-3 events per year from black hole binaries with component masses 10-15 M ⊙ , with a signalto-noise ratio of around 2-3, in each of a network of detectors consisting of GEO, VIRGO and the two LIGOs. It appears that other possible sources, including coalescing neutron stars, are unlikely to be detected by the initial instruments. We also argue that relic gravitational waves may be discovered by the space-based interferometers in the frequency interval 2 × 10 −3 Hz-10 −2 Hz, at the signal-to-noise ratio level around 3.