Working memory capacity is usually measured as the number of stimuli correctly remembered. However, these measures lack precision when assessing individuals with low capacity. This study aimed to create a more precise measure of visuospatial working memory capacity, using intra-level differences in difficulty between items. In two experiments, children aged 4-6 years (N = 97) were tested on a large number of items. Data showed a large variability of difficulty within each level and the factors contributing to this variability were identified. This variability can be used to provide a precise measure of working memory capacity in the lower ranges.