This article used two studies to investigate sign and symbol mediation in children aged 8-11 years. In role play, children exist at one at the same time in objective reality and their representation of reality. We cannot observe their mental representation directly, but the issue of whether signs or symbols mediate early role play is an important theoretical question. The Piagetian perspective is that the symbolic representation develops alongside the sign representation, whereas the psychoanalytic perspective is that symbolic representation of the world precedes sign representation. Work by Venger, DeLoache and Van Oers shows that young children have access to one type of modelling that approximates to sign representation. In this type of model the children can only map relations between the model and the reality that it represents. Models with no direct relation to reality require symbolic mediation, in which children are aware of the model as a representation and explore features of the model itself. Uncertainty provokes the use of symbols or emotional images, and these two studies were designed to assess the effect of uncertainty on the representations children used to solve problems. In the first study, 57 children aged between 8 and 11 were shown a box which lit up red and green lights in an invariant sequence which could only be predicted by memorisation. Children were asked to guess the colour of the next light. Their strategies were classified as "random", "objective symbolisation", "modelling" and "subjective symbolisation". The study showed that many children used sign representations (the modelling strategy) and that the use of symbols as a cognitive tool was uncommon. In the second study 84 children aged between 8 and 11 were asked to match cards thematically. Their strategies were defined as "similarity", "common criterion", "objective symbolisation" or "subjective symbolisation", with "similarity" being the most frequent strategy used. Correlations showed that "common criterion" and "objective symbolisation" (symbolic representation) strategies tended not to be used along with "similarity" and "subjective symbolisation" (sign representation) strategies. Factor analysis revealed two factors that confirmed this analysis. Results show that in familiar tasks children used sign mediation, whereas in more uncertain situations children used symbolic mediation. (Contains 2 figures, 1 table and 1 note.)