Background: Examined or unexamined, all teacher candidates (TCs) bring to their teacher education programs their prior school experience as well as the socio-cultural history of education. This study was one of a series done by this researcher (1994, 2004, 2005, 2006) to find ways to promote reflective practice in teacher education. Purpose: The purpose of this study was to explore ways that teacher educators could help TCs examine their beliefs about teaching and learning so that they could affirm, reject, or modify them as they embarked on their teaching careers. Setting: This research was a component of a yearlong graduate initial teacher certification program for returning adults that combined concurrent coursework with a yearlong internship in a single school. Study Sample: The 115 graduate students included: 67 elementary level TCs (53 female and 14 male); and 47 secondary (27 female and 20 male). Intervention: In this mixed methods study, TCs drew a picture of what for them, as a student, would be an actual and an ideal classroom experience. TCs completed the same 18 question self-report on the learning styles they were depicting in each picture. TCs then listed the qualities of a good, an effective, and an ineffective teacher. These lists were shared with peers in partner and full group discussions. Research Design: Quasi-experimental; Other Quantitative; Qualitative. Control or Comparison Condition: Not relevant to this research. Data Collection and Analysis: The pictures provided a specific stimulus and context from which to consider one's values. The completed self-report learning style questionnaires that used a Likert rating scale were statistically compared for consistency between the actual and ideal pictures. Two expert raters completed content analyses of the pictures and tallies were made of the TC's lists of teacher attributes. Findings: The results of these studies strongly supported the use of the actual and ideal school drawings to determine TCs' learning preferences and beliefs about teaching and learning. TCs expressed learning style preferences that were consistent between their actual and ideal school pictures. No single learning style preference was predominant. Seventeen of the 18 constructs met or exceeded the threshold for = 60% for consistency. The TCs generated three lists of a wide range of qualities: 123 different attributes of a good teacher, 83 of an effective teacher, and 64 of an ineffective teacher. Conclusion: The TCs in this study were able to communicate consistent learning preferences. Their drawn images realistically represented both historical and current beliefs about teaching and learning, which Cuban (1993) said have stayed the same for a century. The research tasks served as a springboard for self-reflection and discussion. The analysis of the pictures helped explain the TCs' frustrations when their own ideal visions were not actualized or allowed within their field placement. They also allowed the TCs to see how the reflections of past practice were limiting their ability to envision alternative ideals. (Contains 3 tables and 6 figures.)