Research suggests that daycare children develop a network of attachment relationships both within and outside of the family and that all attachment relationships are important to the development of children's emotional and social development. In Canada today, approximately two-thirds of mothers with children younger than school age are in the paid labour force, and there are a large number of children being raised in what has been labeled "other-than-mother care" (Scarr, 1984). Based on the presumption that affectionate and angry behaviours on the part of daycare educators would have major implications for the development of the children for whom they care, the aim of the present investigation was to provide the groundwork for identifying what factors were associated with educator warmth and anger in daycare settings.  The conceptual framework for the present study draws upon the ecological models (Bronfenbrenner, 1986) that explicitly acknowledge the multiple levels of environmental influence on the caregiver's behaviour in order to study the correlates of these behaviours. Thus, several categories of variables were used to predict affectionate and angry caregiver behavior. These included educator characteristics, personal resources, the work environment, wages and global quality, and the caregiver's perception of her work.  A multi-method, multi-respondent approach was employed, including researcher observations, educator self-report questionnaires, and objective data collected from directors and from Québec's Office des Services de Garde à l'Enfance. Seventy-eight female educators from 37 centres caring for preschool-aged children were observed in their classrooms for approximately two hours, using two valid and reliable time-sampling outcome measures of affection and anger. Characteristics of their work environments were also recorded. The educators then completed a battery of questionnaires in which they provided information on their background characteristics, personal resources and perceptions of their professional roles and details about their work environments. Directors were asked to complete a brief questionnaire concerning characteristics of their centre and their own background.  The findings from this study suggest that different sets of variables are related to affection and anger. The work environment had a greater relation with caregiver's affectionate behaviour, whereas more internal, negative perceptions were linked to the expression of anger in the classroom. Improvements in the work environment might increase warm interactions directly by liberating the educator's resources in order that she can devote more of her energy towards being affectionate with the children. Changes in the work environment, however, may only decrease the educator's expressions of anger if she perceives herself as being supported in her job.  Though training did not emerge directly as a significant factor in the display of affection or anger, (a) training did have an indirect relation with these behaviours by operating through the work environment, (b) educators who were less sufficiently trained were found to be more angry and less affectionate with the children when another risk factor was present, and (c) under adverse conditions, higher levels of affection and lower levels of anger were noted for higher levels of training, suggesting that training may serve as a protective factor from negative influences