A study was conducted to investigate the respective contributions of personal and social factors (initial competence and imitation of modeled responses) in children's observational learning of conservation. Subjects included 180 first-grade nonconservers from six schools attended by French-speaking middle class pupils. All subjects failed conservation tasks at pretesting; half failed anticipation tasks. Anticipators and nonanticipators were subsequently exposed to one of five pretraining conditions emphasizing (1) generalized imitation of modeled responses different from those suggesting conservation; (2) generalized counter-imitation; or (3) one of three control situations (neutralization pretraining, sensitization period, or no pretraining). Conservation was then demonstrated to all subjects. During the demonstration, one-third of the children were asked to repeat modeled conservation responses after each problem, while an equal number rested during pauses. For the remaining third, no pause interrupted the demonstration. Posttesting was conducted immediately following the demonstration and 4 weeks later. As predicted, performance of anticipators surpassed that of nonanticipators, who showed no progress at all. For anticipators, imitation pretraining provided optimal benefits. Moreover, rehearsal manipulation had no influence on performance. These results were considered in light of a hypothesized sequence of imitation and comprehension processes, the evolution of which is modified by the nonconserver's cognitive preparation and the situational encouragement of imitation. (Author/RH)