How can we best discover the ultimate, evolved functions of endocrine signals within the field of human behavioral endocrinology? Two related premises will guide my proposed answer. First, hormones typically have multiple, simultaneous effects distributed throughout the brain and body, such that in an abstract sense their prototypical function is the coordination of diverse outcomes. Second, coordinated output effects are often evolved, functional responses to specific eliciting conditions that cause increases or decreases in the relevant hormones. If we accept these premises, then a natural way to study hormones is to hypothesize and test how multiple eliciting conditions are mapped into coordinated output effects via hormonal signals. I will call these input-output mappings "theoretical frameworks." As examples, partial theoretical frameworks for gonadal hormones will be proposed, focusing on the signaling roles of testosterone in men and on estradiol and progesterone in women. Recent research on oxytocin in humans will also be considered as an example in which application of the theoretical framework approach could be especially helpful in making functional sense of the diverse array of findings associated with this hormone. The theoretical framework approach is not especially common in the current literature, with many theories having eschewed explicit consideration of input-output mappings in favor of parsimony-based arguments that attempt to find the one main thing that a hormone does with respect to psychology or behavior. I will argue that these parsimony-based models have many shortcomings, and conclude that the construction and testing of theoretical frameworks provides a better means of discovering the evolved functions of human endocrine signals.