This study examined the development of cognitive perspective taking skills and the lack of consistency across perspective taking measures in earlier studies. Four perspective taking measures were administered to 56 children between 4 and 10 years of age under two testing conditions. The high structure condition included multiple presentation of the task to the child, clear task definition, high task salience, clarity of response options, and demonstration of the task. The low structure condition did not include these elements. The correlations between perspective taking tasks in the high structure assessment were higher than the correlations between tasks in the low structure assessment. All measures were related to age and verbal mental age, that is, IQ divided by chronological age. IQ was not related to performance in either condition. These results warranted four conclusions: (1) Cognitive perspective taking is related to age; (2) Different perspective taking skills develop in similar ways; (3) The different contents of perspective taking develop in a manner consistent with a general model of skill development; and (4) The lack of control of task and testing conditions, rather than a lack of coherence of the cognitive perspective taking construct, caused the absence of convergence of measures found in earlier studies. (BC)