The search for indicators of teaching expertise has yielded several models of teacher professional development. These models usually describe teacher development as gradual qualitative shifts in teachers' professional thinking, attitudes, mentality, performance etc. For example, the Fuller and Brown's (1975) model opens teacher professional development in the light of changes in teacher concerns. Another model, elaborated by Berliner (1988), describes professional development as a series of qualitative changes in teachers' professional decision-making capabilities. Yet, a more recent study by Sato and his colleagues (1993) learned differences in beginning and experienced teachers' practical thinking styles. The study showed that experienced teachers' thinking, in comparison with novices, is more fluent and coherent, they get better involved and are better capable of switching from one teaching perspective to another, they see the classroom events in the appropriate context, and have a more advanced level of hypothetical thinking than novice teachers. This paper introduces a research program aimed at studying differences in Estonian teachers' thinking styles. The principal aim of the study is uncovering indicators of teacher professional development by learning differences in teachers' perceptual capabilities as reflected in their comments on classroom events. The adopted research methodology foresaw implementation of four consecutive tasks. Firstly, a procedure for selecting teachers with different and identified levels of professional development as research subjects was needed. The used procedure was based on two principles: (1) only representatives of the two most contrasting teacher experience groups - novice and expert teachers - were targeted; (2) teachers were identified as experts, if they had at least ten years' of working experience at school, a high certification rank, and were recognized as good teachers at their own schools.Secondly, a stimulus material was needed for eliciting reactions reflecting differences in subjects' perceptual capabilities in the fixed conditions. To this end a videotaped lesson of Estonian language in grade seven was used. Before the observation the subjects were instructed to comment aloud all lesson events that caught their attention. The comments were tape-recorded and later transcribed. Thirdly, a reliable methodology was necessary for categorizing and analyzing the subjects' comments. In order to make findings of Estonian study comparable with those of the Japanese study, an attempt was made to reconstruct the content analysis procedures used by Sato et al. However, in the context of Estonian language and culture several measures were necessary for improving the reliability of these procedures. Firstly, the video-recorded lesson used as stimulus material was divided into main instructional events and chrometrized. The content analysis of teachers' comments was carried out at three different levels of generalization: the analysis of comments on (1) teacher and student activities directly related to instructional events as defined by Gagné (1985), (2) teacher and student activities not directly related to instructional events, and (3) general atmosphere or approach to teaching reflecting class teaching and learning activities in the long run. A reading guide for identifying and coding the transcribed teacher comments into categories was prepared for increasing the inter-researcher reliability of categorization. Lastly, the final categorization of teachers' comments was realized on the consensual basis. Fourthly, a methodology for the horizontal analysis of research outputs leading to the definition of indicators of teaching expertise and to the elaboration of ways of their identification was needed. To this end the methods of constant comparison and conceptual framing were used. The study identified many differences between perceptual capabilities of novice and expert teaches susceptible to be considered as indicators of teacher expertise.