One often-used approximation in the study of binary compact objects ( i.e. , black holes and neutron stars ) in general relativity is the instantaneously circular orbit assumption .
This approximation has been used extensively , from the calculation of innermost circular orbits to the construction of initial data for numerical relativity calculations .
While this assumption is inconsistent with generic general relativistic astrophysical inspiral phenomena where the dissipative effects of gravitational radiation cause the separation of the compact objects to decrease in time , it is usually argued that the timescale of this dissipation is much longer than the orbital timescale so that the approximation of circular orbits is valid .
Here , we quantitatively analyze this approximation using a post-Newtonian approach that includes terms up to order ( { Gm / ( rc ^ { 2 } ) } ) ^ { 9 / 2 } 
for non-spinning particles .
By calculating the evolution of equal mass black hole / black hole binary systems starting with circular orbit configurations and comparing them to the more astrophysically relevant quasicircular solutions , we show that a minimum initial separation corresponding to at least 6 ( 3.5 ) orbits before plunge is required in order to bound the detection event loss rate in gravitational wave detectors to < \ > 5 \% ( 20 \% ) .
In addition , we show that the detection event loss rate is > \ > 95 \% for a range of initial separations that include all modern calculations of the innermost circular orbit ( ICO ) .