The role of dreaming in childhood and in adulthood are still equally enigmatic fields yet to be fully explored. However, while there is a consensus at least about the typical content and formal characteristics of adult dream reports, these features are still a matter of debate in the case of young children. Longitudinal developmental laboratory studies concluded that preschoolers' dreams usually depict static images about mostly animals and body states of the dreamer but they basically lack the active representation of the self, human characters, social interactions, dream emotions and motion imagery. Due to methodological arguments these results became the reference points in the literature of developmental dream research, in spite of the significantly different results of numerous recent and relevant studies using extra-laboratory settings. This study aims to establish a methodologically well-controlled and valid way to collect children's dreams for a representative period of time in a familiar home setting to serve as a comparison to the laboratory method. Pre trained parents acted as interviewers in the course of a 6 week-period of dream collection upon morning awakenings. Our results suggest that even preschoolers are likely to represent their own self in an active role (70%) in their mostly kinematic (82%) dream narratives. Their dream reports contain more human, than animal characters (70 and 7% of all dream characters respectively), and social interactions, self-initiated actions, and emotions are usual part of these dreams. These results are rather similar to those of recent extra-laboratory studies, suggesting that methodological issues may strongly interfere with research outcomes especially in the case of preschoolers' dream narratives. We suggest that nighttime awakenings in the laboratory setting could be crucial in understanding the contradictory results of dream studies in case of young children.