From studies over the past 20 years four contrasting hypotheses can be made about the nature of parent-infant communication: (1) mothers and fathers display similar skills to their infants and do not exert a differential influence on their development; (2) fathers are less sympathetic to their infants' level of development and therefore inadvertently stretch the child's development more than mothers; (3) both parents differentially socialize their sons' and daughters' early communicative skills; (4) any apparent differences between parents reflect their expectations about being observed. To examine these hypotheses together, this experiment records the communication of 10 mother-infant and 10 father-infant dyads in two conditions: when an observer was present or absent. The analysis revealed two patterns. Firstly, in keeping with most research on parent-child communication, mothers and fathers both simplified their speech to their infants in similar ways. Secondly, both the structure and function of parental communication showed differences between the two conditions and many of these differences were moderated by interactions between condition and sex of parent or child. The data thus provide more support for the first and fourth hypotheses cited above. It is suggested that analyses of parent-infant interaction should move away from simple assumptions about parental influences upon children's development to consider the subtleties of different parental styles in different settings.