This review considers experiments which have tested triadic balance in POX and POQ structures. It is pointed out that balance is only one bias which affects subjects' responses to these structures. Of the others the most important are agreement and P/O positivity; these three biases act independently of each other and therefore should be considered as separate influences on subjects' responses. In addition, because they act independently, Newcomb's (1968) theory of interpersonal balance cannot be an adequate explanation of their operation. In an attempt to discover the different influences on balance, agreement, and positivity biases, experiments concerned with balance theory are reviewed in three sections: (1) a classification based on experimental task, (2) a classification based on characteristics of the triad, and (3) personality variables. It is concluded that there are different influences on the different biases, and that any attempt to assess their relative strength must consider the influences upon them. Finally, it is suggested that the view of biases as methods of encoding information about social structures should consider that such methods can be varied by subjects' knowledge or assumptions about that social structure.