This qualitative self-study examines decision-making processes in challenging mentorship situations within teacher-education programs. The research identifies criteria for maintaining or terminating teacher-candidate placements, balancing learning potential against risks. Interviews were conducted with program leadership and teacher candidates involved in four critical incidents. Data analysis revealed themes of prioritizing candidate well-being, expectations of success, and interrogating dominant practices. The findings led to developing a Critical Incident Support Model, outlining steps for deciding how to proceed when a teacher candidate has a mentor teacher who may be ineffective or perhaps even causing harm. The model emphasizes gathering data from multiple stakeholders, maintaining open communication, and mobilizing faculty advisors for timely support. Analysis of candidate interviews highlighted the complex emotions experienced during critical incidents. This research contributes to understanding how teacher education programs can effectively mentor candidates by ensuring meaningful mentorship opportunities. The Critical Incident Support Model provides a structured framework for decision-making that prioritizes candidate well-being and learning.