Massive binary stars undergo qualitatively different evolution when the two components are similar in mass ('twins'), and the abundance of twin binaries is therefore important to understanding a wide range of astrophysical phenomena. We reconsider the results of Pinsonneault & Stanek (2006), who argue that a large proportion of binary stars have nearly equal-mass components; we find that their data imply a relatively small number of such 'twins.' We argue that samples of double-lined spectroscopic binaries are biased towards systems with nearly equal-brightness components. We present a Monte-Carlo model of this bias, which simultaneously explains the abundance of twins in the unevolved binaries of Pinsonneault & Stanek (2006), and the lack of twins in their evolved systems. After accounting for the bias, we find that their observed mass ratios may be consistent with a variety of intrinsic distributions, including either a flat distribution or a Salpeter distribution. We conclude that the observed overabundance of twins in Pinsonneault & Stanek (2006) does not reflect the true population of binaries, and we briefly discuss the astrophysical implications of the lack of twins.