A great deal of work has been done on the theory of mass loss and evolution in close binaries, and numerous individual systems have been discussed in this connection, but the general question of the binary frequency of O-stars, and in particular, the initial binary mass ratio frequency or distribution of secondary masses, has not been completely answered. In general, we know that about half of all O-type stars are binaries; the most recent determination by Conti, Leep and Lorre (1977) found 58% of their sample to be certain or probable binaries. However, many of these stars were judged to be variable on the basis of only a few spectra from different sources, and therefore require further study. Another point to be examined concerns the binaries with available orbits: two thirds of these are double line systems. Figure 1 shows a plot of the semi-amplitude versus orbital period for all known systems, along with some theoretical curves for different mass ratios. Not only is the lack of single line systems obvious, but low amplitude systems are almost completely missing. This would appear to be only an observational selection effect, although it is to be noted that low amplitude double line Wolf-Rayet systems have been detected. If the effect is real, it implies that O-type binaries with mass ratios (m1/m2) greater than about three do not exist.