The general hypothesis concerning the development of dyadic interaction ascribed to here is that during the first six weeks after birth the infant is insufficiently equipped for active participation in en face interaction and acts relatively independently of his mother. Due to the maturation of underlying neural mechanisms at the age of about two months, a number of crucial transformations occur in postural, motor and visual functions (see PRECHTL, 1984). These developmental changes enable the infant to become an increasingly active partner in en face interaction: a two-way process during which both mother and infant relate the timing of their behaviour to that of the other. Six healthy mother-infant pairs participated in this longitudinal, home-based study which made use of video equipment to record a 15 min interaction session at 3, 6, 9, 12, 15 and 21 weeks of age. The infant behaviours selected for study were: posture, movement, looking, smile, whimper, grunt, "pleasure" vocalization, yawn, grasping hands in mid-line, hand-mouth contact. Maternal behaviours selected were: looking, body movement, head movement, touching, speech, other vocalizations and sound effects, moving infant's limbs. The aim of the study was to provide a quantitative description of the changes in the sequential dependence of behaviour within the mother-infant pairs. To this end information theoretical statistics as elaborated by van den BERCKEN (1979) and van den BERCKEN & COOLS (1980) were applied making use of analysis of variance terminology. The methodological problem of taking into account the constraint stemming from each partner's own previous behaviour (auto-correlation) when making inferences about ongoing behaviour between the two partners (cross-correlation) was removed by identifying and successfully controlling the various sources of influences affecting behaviour during dyadic interaction. The data were addressed to the following questions: 1. Was there a change in the use the infants made of their behavioural repertoire (individual variability of the infant) during the age-range studied? 2. Was there a change in the use the mothers made of their behavioural repertoire (individual variability of the mother)? 3. Was there a change in the constraint on the infant's current behaviour stemming from his previous behaviour, i.e. did the amount of variance explained by the exclusive influence of the infant's own preceding behaviour on its current behaviour (auto-covariability) increase during the age-range studied? 4. Was there a change in the constraint on the infant's current behaviour stemming from his mother's previous behaviour, i.e. did the amount of variance explained by the exclusive influence of the mother's previous behaviour (cross-covariability) increase during the age-range studied? 5. Was there a change in the constraint on the infant's current behaviour stemming only from the combined effect of his own and his mother's previous behaviour, i.e. did the so-called synergic covariability effect become stronger during the age-range studied which reflected increasing mutual dependence between the behaviour of the two partners? For both mothers and infants an increase was found in individual variability while no changes were found in the measures of auto- and cross-covariability. The measure of synergic covariability became increasingly stronger from 12 weeks onwards. These results were interpreted as follows: over the age-range studied infant and mother showed a growing mutual dependence (stronger synergic covariability) with increasing use of their own repertoire (increasing individual variabilities) while the infant's behaviour neither became less dependant on his own previous behaviour (auto-covariability) nor more dependant on the mother's previous behaviour (cross-covariability). Thus a quantitative confirmation of the general hypothesis stated at the onset was provided. Moreover, information theoretical statistics have proved a satisfactory method by means of which dyadic interaction between mother and infant can be described quantitatively.