This study assessed estimation skills on both static and transformation (conservation) tasks on numerosities ranging from 3 to 16, and included addition and subtraction trials to control for response bias. A total of 148 four- and five-year-old children estimated the number of balls of yarn sewn on elastic straps, and participated in 1 addition, 1 subtraction, and 2 conservation trials at each numerosity of the transformation task. The children were classified into three groups based on their performance: (1) conservers, who gave at least one adequate explanation on a conservation trial; (2) transitionals, who did not explain adequately, yet quickly gave at least 90 percent correct judgements; and (3) nonconservers, who could not adequately explain the result and gave incorrect judgements. Contrary to the hypothesis, conservers were no more accurate than transitionals in number conservation tasks. (MDM)