A total of 72 kindergarteners received eight conservation of number trials which varied in the degree of perceptual support for one-to-one correspondence (four levels) and type of stimuli (toy animals or corks). A between-subjects variable was the method of presentation (standard conservation presentation, a partially fixed array, or a fixed array). There was a strong order effect in the direction of increasing conservation over the eight trials. There was some evidence that the standard conservation presentation was easier than the other two presentations. The degree of perceptual support and type of stimuli had no effect. Results were discussed in terms of the competence-performance distinction and implications for assessment and training studies. (Author/CS)